Making the most of your host: the Metrosideros-feeding psyllids (Hemiptera, Psylloidea) of the Hawaiian Islands

Abstract The Hawaiian psyllids (Psylloidea, Triozidae) feeding on Metrosideros (Myrtaceae) constitute a remarkable radiation of more than 35 species. This monophyletic group has diversified on a single, highly polymorphic host plant species, Metrosideros polymorpha. Eleven Metrosideros-feeding species included in the Insects of Hawaii by Zimmerman are redescribed, and an additional 25 new species are described. Contrary to previous classifications that placed the Metrosideros-feeders in two genera, Trioza Foerster, 1848 and Kuwayama Crawford, 1911, all 36 named species are placed in Pariaconus Enderlein, 1926; and the relationship of this genus to other Pacific taxa within the family Triozidae, and other Austro-Pacific taxa feeding on host plants in Myrtaceae is clarified. The processes of diversification in Pariaconus include shifts in galling habit, geographic isolation within and between islands, and preferences for different morphotypes of the host plant. Four species groups are recognized: the bicoloratus and minutus groups are free-living or form pit galls, and together with the kamua group (composing all of the Kauai species) form a basal assemblage; the more derived closed gall species in the ohialoha group are found on all major islands except Kauai. The diversification of Pariaconus has likely occurred over several million years. Within island diversification is exemplified in the kamua group, and within species variation in the ohialoha group, but species discovery rates suggest this radiation remains undersampled. Mitochondrial DNA barcodes are provided for 28 of the 36 species. Genetic divergence, intraspecific genetic structure, and parallel evolution of different galling biologies and morphological traits are discussed within a phylogenetic framework. Outgroup analysis for the genus Pariaconus and ancestral character state reconstruction suggest pit-galling may be the ancestral state, and the closest outgroups are Palaearctic-Australasian taxa rather than other Pacific Metrosideros-feeders.


Introduction
The Hawaiian Islands are renowned for exemplary species radiations (e.g. Magnacca and Danforth 2006, Rubinoff 2008, Givnish et al. 2009, Lerner et al. 2011, Bennett and O'Grady 2012, Goodman et al. 2014, Magnacca and Price 2015, as well as extraordinary and varied processes of species diversification and evolution (Rivera et al. 2002, Mullen et al. 2007, Rubinoff and Schmitz 2010, Wessel et al. 2013, Gillespie 2016. The primary characteristics of the Hawaiian archipelago that are considered important in speciation processes are: a) multiple islands, b) each island with a geographically concentrated heterogeneity of habitats, c) islands occurring in various degrees of isolation and age in relation to one another, and d) the extreme remoteness of the archipelago. Archipelagos with heterogeneity of habitats and varying geological ages are also found in other island systems, but the extreme isolation of the Hawaiian Islands is unique and may ultimately influence patterns of diversification in ways not found elsewhere (Gillespie et al. 2012, Shaw andGillespie 2016).
The Hawaiian Islands are home to several endemic psyllid lineages that represent multiple independent colonizations, yet an enduring puzzle is why all of these lineages are from a single psyllid family. The native psyllid fauna is composed entirely of species from the family Triozidae; a single Liviidae specimen (collected in 1925) of an undetermined Paurocephala Crawford, 1913 species (Crawford 1927) remains unconfirmed (Zimmerman 1948). Adjacent continental and Pacific island regions are home to all eight psyllid families (notably Aphalaridae, Carsidaridae, and Psyllidae; Hodkinson 1983, 1986, 1988, Yang and Raman 2007, including the Marquesas and Austral Islands to the south of the Hawaiian Islands. The question therefore remains: why have none of these other psyllid families colonized the Hawaiian archipelago? Perhaps triozid psyllids are somehow more successful colonizers of remote islands. Other Pacific archipelagos are located within 1000 km distance of habitable landmasses (via many small islands), whereas the Hawaiian Islands are more than 3000 km from another landmass, which makes dispersal ability a critical factor (Hembry et al. 2013). However, it is difficult to reason how dispersal ability alone could discriminate between psyllid families. The colonization of remote landmasses containing few familiar host plants could be facilitated by an ability to survive on suboptimal or unfamiliar hosts Percy 2008, Percy 2011), and although the majority of psyllids are mono-or oligophagous (feeding on one or a few plant species within the same genus), the family Triozidae includes the largest number of psyllid species exhibiting atypically broad host preference (Ouvrard et al. 2015). An ancestral preadaptation to expanded host ranges (Janz et al. 2006, Janz andNylin 2008) may be an advantage allowing successful colonization that may then cycle back to specialization given ecological opportunity (e.g. vacant niches, proximity of alternate hosts) (Percy et al. 2004). This scenario could explain both the imbalance in colonization potential of the different psyllid families and the observed pattern of within archipelago host specialization after establishment. There are also examples from introduced species of host range expansions post colonization of a new region, e.g. Trioza eugeniae Froggatt, 1901, which is associated with Syzygium (Myrtaceae) in its native Australian range, but in California where it is introduced it occurs on cultivated Metrosideros (Percy et al. 2012).
Many of the Hawaiian triozid species have been ascribed to endemic genera rather than placed within more widespread and established generic groupings; exceptions include taxa placed in Kuwayama Crawford, 1911 andTrioza Foerster, 1848, two highly artificial genera found in old and new world regions (Hodkinson 1983(Hodkinson , 1986(Hodkinson , 1988. The practice of erecting endemic Hawaiian genera does, however, often reflect unusual morphology as well as ambiguous affinities to genera elsewhere. The absence of such conspicuous morphological characters distinguishing Metrosideros-feeding species in the Hawaiian Islands is notable, and this is partly responsible for the placement of these taxa in Kuwayama and Trioza. Trioza, in particular, is a large, poorly defined, polyphyletic genus into which many disparate species have been placed (Burckhardt andOuvrard 2012, Ouvrard et al. 2015). Kuwayama is also considered problematic and currently includes unrelated taxa from the American continent and Hawaiian Islands. Although both Enderlein (1926) and Crawford (1918) recognized the artifice of placing Hawaiian taxa in Kuwayama, it was Enderlein (1926) who removed three of the Metrosideros-feeding species to a separate, endemic genus, Pariaconus Enderlein, 1926; not based on unique differences, but rather on an absence of distinct affinities with other taxa in Kuwayama (Enderlein 1926). Despite the absence of distinct morphological synapomorphies, endemic generic status for the Metrosideros-feeding species is clearly supported by this study. Pariaconus is not close to the type species of Trioza, Trioza urticae (Linné, 1758), nor to the type species of Kuwayama, Kuwayama medicaginis (Crawford, 1910). However, closer taxonomic affinities outside the Hawaiian Islands remain to be clarified with yet further outgroup sampling.
All Hawaiian psyllid lineages appear to exhibit high host specificity that has been conserved during in situ diversification (e.g. Pariaconus on Metrosideros (Myrtaceae), Swezeyana Caldwell, 1940 on Planchonella (Sapotaceae), Hevaheva Kirkaldy, 1902 on Melicope (Rutaceae), Megatrioza Crawford, 1915 on Pritchardia (Arecaceae)) (Zimmerman 1948, Nishida et al. 1980, Uchida and Beardsley 1988. Pariaconus is an endemic Hawaiian genus with more than 35 species found on a single host plant (Metrosideros polymorpha) (Table 1) and species discovery rates suggest this number will continue to increase. This pattern of within host diversification is atypical for psyllids, as well as for other phytophagous insects (Joy and Crespi 2007). The rarity of speciose radiations on a single host plant is thought to be due, in part, to processes such as competitive exclusion between closely related species (Percy 2003b, 2011, Després and Cherif 2004, Gold et al. 2009). A much more common process associated with speciation in phytophagous insects, including psyllids, is switching to different host plants (Tilmon 2008, and chapters therein, Ouvrard et al. 2015), and a high degree of host specialization in psyllids appears to constrain most switching to closely related plant species with much rarer saltationary leaps to unfamiliar hosts (e.g. Burckhardt and Basset 2000, Percy et al. 2004, Ouvrard et al. 2015. The question is therefore whether diversification in Pariaconus evolved via atypical evolutionary processes, or whether typical evolutionary pathways have been co-opted in atypical ways. In order to understand how the Pariaconus radiation has evolved and persisted on M. polymorpha, it is crucial to appreciate the ecological background over which the speciation processes have played out. Except in the very driest regions, M. polymorpha is the dominant native shrub in the Hawaiian Islands (Mueller-Dombois 1987, Dawson and Stemmermann 1999. There are currently eight M. polymorpha varieties recognized, but these inadequately represent the complex morphological variation observed within and between islands , Harbaugh et al. 2009, Wright and Ranker 2010, Stacy et al. 2014. In growth habit, the variation includes statuesque trees (~20 m) and low growing shrubs (< 1 m). Habitats may be wet or dry forest, bogs, or almost soil-less lava flows a few years old (Stacy et al. 2014(Stacy et al. , 2016. Studies of the morphological variation (e.g. leaf pubescence and thickness), including common garden and genetic studies, have concluded that the diversity of phenotypes is controlled by complex genome-environment interactions, as well as population dynamics (Cordell et al. 1998, Gruner et al. 2005, Wright and Ranker 2010, Stacy et al. 2014. Indeed, the presence of different morphotypes growing side by side in the same environment but exhibiting little or no genetic variation for neutral markers is one of the most striking features of this polymorphic plant species , Wright and Ranker 2010, Stacy et al. 2016. The patterns of diversity observed in Pariaconus would not be markedly different from other radiations (e.g. those found in the Atlantic island archipelago of the Canary Islands; Percy 2003aPercy , 2003b if the variation found in M. polymorpha was represented by 20 or more distinct species with speciation occurring primarily via a process of switching between hosts. The Canary Islands, although less remote, share many attributes with the Hawaiian Islands, e.g. five major islands formed in an age progressive series over a volcanic hot spot, with similar degrees of habitat heterogeneity and interisland distances (Neall andTrewick 2008, Bogaard 2013). In the Hawaiian Islands, the high degree of polymorphism in M. polymorpha may be promoting psyllid speciation processes akin to the effect of multiple host plants. However, understanding the diversification processes is further complicated by shifts in galling biology, and shifts to galling different plant organs.
When considering the origins of Hawaiian Pariaconus, Crawford (1918) thought "the original immigrant to be one inhabiting leaf galls", and although originally describing different Pariaconus taxa in separate genera, he also stated, "as new species evolved from this, some have retained the gall-making habit … others have taken to living free in the nymphal stages, while still others have gone off to other plants making leaf galls or living free", in other words he contemplated the scenario that all the Metrosideros-feeders, and even all Hawaiian Triozidae could be a monophyletic group. This illustrates Crawford's ambiguous approach to the concept of monophyly. He apparently saw no conflict in simultaneously considering evolutionary shifts in situ within the Hawaiian Islands, but at the same time ascribing these island species to different genera based on superficial morphological affiliations with continental groups. Indeed, Crawford noted, when assigning three newly described Hawaiian Metrosideros-feeding species to Kuwayama, that "it seems certain that the species placed in this genus do not represent a common origin at all, but independent or parallel evolution toward the same end" (Crawford, 1918). He goes on, "the three species seem almost certainly to have been derived from some Trioza species, probably T. ohiacola [P. ohiacola], or an ancestral type proceeding it". He is apparently confirming here that he views taxonomic groupings as artificial and more for convenience than necessarily representative of evolutionary origins.
Multiple shifts in galling biology on the same plant species have rarely been documented. One of the few known systems is the gall-inducing Asphondylia Loew, 1850 (Cecidomyiidae) flies found on Larrea tridentata (Joy and Crespi 2007) where diversification has involved numerous shifts between different plant organs (leaves, buds, flowers, and stems) of the same host-plant species, thereby ecologically partitioning the host plant, with additional temporal as well as spatial separation. In a study of oak-gall wasps, Cook et al. (2002) found speciation more likely to involve shifts to galling different plant organs if the host oaks were more closely related (within taxonomic sections) than when host shifts involved oaks in different sections. Another study looking at Eurosta Loew, 1873 (Tephritidae) gall flies on Solidago where sympatric speciation has involved shifting between leaves and stems has suggested gall makers are more likely to speciate sympatrically than non-gall makers (Craig et al. 1994). These findings are consistent with patterns observed in the Metrosideros-psyllid system and thus help to frame hypotheses that may explain diversity in Hawaiian Pariaconus. However, analysis of the processes of speciation in Pariaconus may be additionally complicated by the underlying complexity inherent in the mosaic of plant phenotypes, and the frequent co-occurrence of alternate Metrosideros morphotypes.
The objectives of this study are, a) to describe the diversity of species feeding on Metrosideros in the Hawaiian Islands, b) to identify patterns of diversity within and between islands across the archipelago, c) to test putative outgroup affiliations, and d) to better understand the origins and evolution of Pariaconus.

Sampling
Over 500 specimens from museum and field collections were examined (Tables 2-3 (Oahu, Maui). Thirtyeight outgroup taxa were sampled from Hawaiian and other island and continental regions. Adults and immatures were collected in the field into 95% ethanol. To associate adults with immature biologies, immatures were sampled whenever possible (e.g. removed from galls, and in some cases adults were reared from galls by placing cut branches in plastic bags). Specimens were preserved in 95% ethanol and stored in -20°C for morphological and DNA analyses. Pariaconus type and other material examined (m -male, f -female, i -immature) with GenBank numbers for cytochrome oxidase one (COI) and cytochrome B (cytB).  Kuwayama minutura (Caldwell, 1940)   Trioza malloticola (Crawford, 1928)

Molecular analysis
The molecular analysis includes 537 individuals, 479 in Pariaconus and 58 in outgroup taxa. To confirm adult/immature/gall types and taxon group associations, DNA barcodes were sequenced from two mitochondrial gene regions, cytochrome oxidase I (COI), and cytochrome B (cytB). Protocols for DNA extraction, PCR, and sequencing follow those described in Percy (2003b), except for annealing temperature, 56°C for cytB versus 50°C for COI; PCR primers for COI and cytB respectively are given in Simon et al. (1994), and Timmermans et al. (2010). Distance neighbour-joining (NJ) analyses were performed using program PAUP* (Swofford 2003), and maximum likelihood (ML) with RAxML (v.8.2.4) on CIPRES (Miller et al. 2010, Stamatakis 2014. Genetic distances reported use uncorrected (p) distances estimated in PAUP* (Swofford 2003). The molecular phylogenetic analyses reported here are intended as confirmation of morphological species concepts, as well as to compare genetic variation with geographic and morphological variation. Further phylogenetic analyses will focus on assessing different rates of evolution in this group. Genomic DNA was extracted from whole specimens, adult or immature, and post-DNA extraction voucher specimens retained in ethanol or slide mounted. The DNA sequences are deposited in GenBank (Tables 2-3). Thirty-seven outgroup taxa in the family Triozidae were included in the phylogenetic analyses, with a focus on taxa putatively ancestral to the Hawaiian Metrosideros-feeders, including Metrosideros-feeders from French Polynesia and New Zealand, other Myrtaceae-feeders from the Pacific-Australasian region, taxa producing galls from continental regions on both sides of the Pacific, the type species of Trioza, and some of the other endemic genera in the Hawaiian Islands. To provide a root for Triozidae, Mesohomotoma hibisci (Froggatt, 1901) To interpret the putative ancestral biology (i.e. galling versus non-galling) for Pariaconus, an ancestral character state analysis was performed using Mesquite (v.3.11) (Madison and Madison 2016) with both parsimony (unordered) and maximum likelihood (Mk1 model) reconstructions. A single three state character, states: free-living, open galling, closed galling, plus "unknown", was traced onto the majority-rule consensus topology of the maximum likelihood analysis using a single representative for each Pariaconus species, and the two Trioza species (that are strongly supported as the nearest outgroups) either as monophyletic or paraphyletic.

Morphological analysis
Ethanol preserved material was cleared in 10% potassium hydroxide followed by clove oil and slide mounted in Canada balsam as described in Hodkinson and White (1979). In some instances, 1 st -2 nd instar immatures were mounted directly into Euparal from 95% ethanol without clearing. Post-DNA extraction voucher specimens were either placed directly into clove oil from 100% ethanol, or were first cleared in 10% po-tassium hydroxide. DNA voucher specimens not cleared before mounting often retain red pigmented ocular tissue after DNA extraction (visible in some Figures). Morphological terminology follows Hodkinson and White (1979), Hollis (1984), White and Hodkinson (1985), and Percy (2003a). Type and other material is deposited in the Natural History Museum, London, UK (BMNH), and in the Bishop Museum, Honolulu, USA (BPBM) (Tables 2-3).
As there is a large degree of size variation in some Pariaconus species (particularly ohialoha group), averages (av.) across all individuals measured are used in descriptions and keys with the measured range given in Tables 4-7. Male parameres are illustrated as outlines (without setation added) because the shape is in most cases sufficient for differentiation, where shape is similar between species, the setation is relatively uniform, and not as informative as other characters illustrated. Suppl. material 1 illustrates adult characters and measurements referred to in the text. Abbreviations used in the descriptions and Tables 4-7 are as follows (all measurements are recorded in mm): Adults: WL, fore wing length; WW, fore wing width; HW, head width; AL, antennal length; GP, genal process length; PB, distal proboscis segment length; WL:WW, ratio fore wing length:width; CUR, ratio fore wing cell cu 1 width:height; MR, ratio fore wing cell m 2 width:height; HW:VW, ratio head width:vertex width; VL:GP, ratio vertex length:genal process length; VW:VL, ratio vertex width:length; AL:HW ratio antennal length:head width; HW:HT ratio head width:hind tibia length. Adult male terminalia: MP, proctiger length; PL, paramere length; AEL, distal aedeagus segment length; PL:HW, ratio paramere length:head width; MP:PL, ratio proctiger length:paramere length; PL:AEL, ratio paramere length:distal aedeagus segment length; AEL:AELH, ratio distal aedeagus segment length:aedeagus apical head length; PL:SH, ratio paramere length:subgenital plate height. Adult female terminalia: FP, proctiger length; FSP, subgenital plate length; RL, anal ring length; OVH, ovipositor valvulae dorsalis height; EL, egg length; EW, egg width; FP:RL, ratio female proctiger length:anal ring length; FP:HW, ratio female proctiger length:head width; FP:SP: ratio female proctiger length:subgenital plate length; EL:EW, ratio egg length:egg width. Immatures: BL, body length; BW, body width; WPL, fore wing pad length; CPL, caudal plate length; CPW, caudal plate width; RW, circumanal ring width; HW, head width; AL, antennal length; BL:BW ratio body length:width; HW:AL ratio head width:antennal length; CPW:RW ratio caudal plate width:circumanal ring width.
Data resources. The collections and specimen data underpinning the analyses reported in this paper are deposited in the Natural History Museum Data Portal as Diana Percy (2017)

Taxonomic placement of Pariaconus and outgroup analysis
Combined analysis of the two mitochondrial regions provides two robustly supported key results: a) confirmation of the monophyly of the genus Pariaconus, b) strong support for a sister relationship with pit-galling species from Asia-Australasia as the closest known relatives of Pariaconus (Fig. 1). There is strong support for the monophyly of Pariaconus (i.e. both taxa previously placed in Trioza and Kuwayama) (ML: 99%, NJ: 100%), and robust support for an ancestral outgroup (ML: 98%, NJ: 98%), which may be suggestive of a leaf pit-galling ancestry (see results of ancestral character state reconstruction below). The most unexpected discovery is the identity of the closest outgroup taxa, which are not among other Pacific Metrosideros -or Myrtaceae-feeders, but include Trioza remota Foerster, 1848, a Palaearctic species making barely noticeable pit galls on the leaves of oaks (Quercus spp.; Fagaceae), and an undescribed species from Australia making deep pit galls on Banksia (Proteaceae) leaves. Other Myrtaceaefeeders, including those on Metrosideros, from the Australasian and Pacific regions do not cluster close to Hawaiian Pariaconus. The French Polynesian taxa on Metrosideros are a separate radiation of species related to Myrtaceae-feeders from Asia and Australia (Trioza outeiensis Yang, 1984, T. eugeniae Froggatt, 1901 and also cluster with taxa galling Eucalyptus and other Myrtaceae (Schedotrioza spp., T. obunca Fang & Yang, 1986, T. vitiensis Kirkaldy, 1907. Trioza eugeniae is an Australian Myrtaceae-feeder associated with Syzygium in its native Australian range, but in its introduced range in California it occurs on cultivated Metrosideros (Percy et al. 2012); this species was therefore considered another putative outgroup for Pariaconus. The phylogenetic analyses revealed it is not related to Pariaconus, but rather groups with the other Pacific-Australasian Myrtaceae-feeders described above. Some doubts about the correct identity of the Californian specimens came to light during this study. Specimens of introduced T. eugeniae sampled from three localities in California are in fact closer to a species described from New Zealand, Trioza adventicia Tuthill, 1952 (specimens supplied by Pam Dale, Plant Protection Centre, MAF, Auckland) which feeds on Angophora floribunda, Syzygium smithii, and S. paniculatum (all Myrtaceae) in New Zealand (Martoni et al. 2016, Pam Dale pers. comm.). Tuthill (1952) recognized T. adventicia as introduced to New Zealand, and based on a morphological examination as well as the molecular analyses, both the Californian and New Zealand specimens sampled here are likely the same species. Despite Tuthill (1952) examining the type material of T. eugeniae when he described T. adventicia, he mentioned only one distinguishing character between them, apical metatibial spurs 2+1 (T. adventicia) versus 3+1 (T. eugeniae). I have examined six specimens (3 male, 3 female) of T. eugeniae sourced from a single California population, and six specimens (5 male, 1 female) of T. adventicia from a single New Zealand population. The California material has four individuals with 2+1 spurs, one with 3+1, and one with both 2+1 and 3+1 (on right and left metatibia). The New Zealand material has five individuals with 2+1 and another with 2+1 and 3+1 (on right and left metatibia). There appear to be no distinguishing morphological characters to separate the Californian and New Zealand material. In conclusion, Californian and New Zealand specimens are likely the same species and may have been introduced from Australia over similar time periods, and possibly from a similar source area in Australia (based on minimal genetic distance). I have not examined sufficient material of Australian T. eugeniae, or type material of either species, which would be needed to confirm synonymization of T. adventicia with T. eugeniae, but molecular divergence between the Australian, and the Californian/New Zealand specimens is well within typical intraspecific distances for psyllids (Percy 2003b, Taylor et al. 2016. Froggatt (1901) mentions different forms of T. eugeniae, which he infers may be different species, suggesting more extensive sampling within Australia is needed to clarify the taxonomy of T. eugeniae. The illustration Froggatt (1901) provides of the male genitalia of T. eugeniae is dissimilar to that of the Californian and New Zealand material I have examined, and as Tuthill was convinced, based on his examination of T. eugeniae type material, of the distinctiveness of T. adventicia, this issue is here flagged as needing further investigation.

Ancestral character state reconstruction: a galling or non-galling origin?
Parsimony reconstruction analysis resolved the ancestral state for Pariaconus as open galling (e.g. pit/cup gallers) with four steps between states within Pariaconus, but character consistency and retention scores were relatively low (CI: 0.5, RI 0.71). The maximum likelihood analysis also recovered open galling as the ancestral state but the margin in proportional likelihoods between open galling and free living is not large (Pariaconus root node with pit galling outgroup taxa monophyletic: 0.52 open galling versus 0.44 free living, and with pit galling outgroup taxa paraphyletic: 0.65 open galling versus 0.32 free living) (see inset in Fig. 1). In all analyses, closed galling is derived within Pariaconus with no reversals. The ancestral trait analysis results remain somewhat equivocal due to the number of unknown biologies within Pariaconus, as well as the lack of firm resolution at the base of the Pariaconus radiation. Finally, lack of knowledge of ancestors and tree topology beyond the two outgroup pit gallers hinders confidence in resolving the question of the original biology, and therefore the results remain only suggestive of the ancestral state and potential evolutionary transitions in galling biology.

Divergence within Pariaconus
Maximum inter-specific molecular divergence within Pariaconus as a whole (across all four species groups) is 16.9%. This can be compared with divergence within Swezeyana and Hevaheva, which, although not as comprehensively sampled, provide maximum divergence estimates of 18% and 17.5% respectively. These comparative divergences suggest Swezeyana and Hevaheva may be marginally older genera that established earlier in the Hawaiian Islands. Mitochondrial rates of divergence vary across different organismal lineages, even among insects (Shapiro et al. 2006, Magnacca and Danforth 2007, Goodman et al. 2012, Haines et al. 2014, and rate heterogeneity has been reported within psyllid lineages (e.g. Percy et al. 2004). However, assuming a mitochondrial molecular clock can still provide a reasonable range of ages in many cases (Papadopoulou et al. 2010, Wessel et al. 2013, Parmakelis et al. 2015. If we assume a range of mtDNA rates between 2.5-5% divergence per million years, then the length of time Pariaconus has been diversifying in the Hawaiian Islands is likely between 3.5-6.5 Myr. This range is also consistent with estimates determined from observed divergence and island age. For instance, if the maximum divergence found in Pariaconus (16.9%) is a product of diversification since the emergence of Kauai (~5 Myr), then divergence rate is estimated at ~3.4% per million years, however, the ohialoha group, which only occurs on the younger islands (with island ages of ~3-3.5 Myr and younger) would suggest that the divergence rate is at least as high as 4.5-5.3% per million years.
Based on morphological and molecular analyses, four species groups are recognized (Figs 1-2). The bicoloratus and minutus groups are free-living or form pit galls (formerly Kuwayama spp.). These taxa are challenging because they include most of the small sized species, which are often less abundant than those in the ohialoha group; and they are often more difficult to detect in the field, particularly the free-living immatures, because there are few or no visible indications of presence on the plant. In addition, when pit galls are shallow they are relatively inconspicuous compared to enclosed galls.
The bicoloratus group is one of the oldest, and possibly the ancestral group in the Hawaiian radiation, but interestingly the greatest extant species diversity in this group is found on the youngest island of Hawaii, although subfossil remains closely resembling extant bicoloratus-type immatures have recently been found on Kauai (Nicholas Porch pers. comm.). The kamua and minutus groups are also likely older than the ohialoha group, but currently the minutus group is only known from Maui and Hawaii. The kamua group is the only single island lineage and includes all of the known Kauai species. The range of morphologies and galling biologies, and the restriction to the oldest extant island of Kauai, also make this group a plausible ancestral group in the radiation. However, of the four larger islands (Kauai, Oahu, Maui, Hawaii), Kauai is the least well sampled with more collecting needed to determine both the habits and variety of galling biologies, as well as the potential presence of bicoloratus and minutus group taxa on Kauai. The ohialoha group includes all of the closed galling species not in the kamua group, and is found on all islands except Kauai; it appears to be the most derived group and exhibits dynamic patterns of variation suggestive of ongoing speciation processes.
Recognized morphological forms provide information about the extent and distribution of morphological variation within, in some cases, relatively broad species concepts, and a number of the more distinct forms may eventually require recognition at species level.

Divergence within the bicoloratus group
There are 12 species in the bicoloratus group and maximum interspecific divergence is 14.4% (maximum intraspecific divergence 10%). This is the only species group not resolved as monophyletic, rather it constitutes a basal grade of taxa likely representing early divergence in Pariaconus (Figs 1-2). The recent discovery of subfossils on Kauai from this group (Nicholas Porch pers. comm.) also supports its ancestral position in the radiation of the genus, although the absence thus far of any extant species from Kauai remains puzzling. There is a subgroup within the bicoloratus group, namely the "bicoloratus species complex", which encompasses complex patterns of morphological and genetic variation not easily interpreted, including two taxa not resolved as monophyletic in either ML or NJ analyses (P. hina, P. wyvernus). Notably, many species are only known from one or few localities, which contrasts with the widespread distributions of many of the closed gallers in the ohialoha group (Figs 53-55)

Divergence within the minutus group
There are two species in the minutus group, and both are well represented in the molecular dataset with 9.2% maximum interspecific divergence (maximum intraspecific divergence 9.1%). Within group divergence appears modest compared to other species groups, and currently most of the genetic divergence is found within P. minutus on Hawaii (Fig. 2), however, the species on Maui, P. gibbosus, is likely undersampled.

Divergence within the kamua group
There are 10 species in the kamua group, with maximum interspecific divergence 14.4%, but molecular data are only available for five species and insufficient individuals Figure 2. Best maximum likelihood topology (-lnL = -33769.25, combined COI and cytB data using RAxML with 1000 bootstrap replicates). Included are 432 unique haplotypes (381 within Pariaconus and 51 representing outgroups) recovered from 537 individuals sampled. The four recognized species groups within Pariaconus are indicated as well as the "bicoloratus species complex" and 28 described species. Moderate to strong support for nodes within Pariaconus is indicated. Dotted lines indicate species either with only one individual sampled or several individuals with only one unique haplotype. Dashed lines indicate two species (P. hina, P. wyvernus) for which intraspecific haplotypes were not recovered as monophyletic, both species are within the "bicoloratus species complex". Two taxa (P. ohiacola, P. caulicalix) are recovered as topologically monophyletic (in both ML and NJ analysis) but without bootstrap support. Two species are recovered as sister taxa in NJ analysis (P. hawaiiensis, P. pyramidalis), but here the ML analysis places one (P. pyramidalis) on a long branch nested within the other (P. hawaiiensis), reflecting a likely evolutionary gall shift in situ on Hawaii. All other taxa, including recognized intraspecific forms and variations, are reasonably to strongly supported as monophyletic (80-100% support). Maximum genetic distance within species groups and maximum intraspecific genetic distances (p-distance in PAUP*) are shown in parenthesis for all taxa with more than one haplotype. were sampled to gauge typical intraspecific divergence, except perhaps in P. caulicalix with 5.2% intraspecific divergence. The kamua group exemplifies within island diversification encompassing a variety of biologies: closed and open galls, and a large range of body sizes. All sampled species were collected within a small geographic area on Kauai (Fig. 54) emphasising the co-occurrence of taxa and the likely discovery of more diversity with broader geographic sampling on Kauai.

Divergence within the ohialoha group
There are 12 species in the ohialoha group; molecular data are presented for 10 species and maximum interspecific divergence is 15.8% (maximum intraspecific divergence 11%). Slightly greater genetic distances in the younger ohialoha group are likely a result of more comprehensive sampling but could also suggest an accelerated molecular divergence rate related to a shift to the closed galling biology. All species, with the exception of P. ohiacola, are well supported, but there is a notable lack of support at the base of the ohialoha group suggesting a rapid radiation occurred after a shift to the closed galling biology (Figs 1-2); subsequent shifts to galling different plant parts then promoted further diversification within the group. Notably, a strongly supported sister relationship between the stem/bud galler, P. hawaiiensis, and cone leaf galler, P. pyramidalis, on Hawaii, provides support for in situ (within island) galling shifts. A sister relationship between a stem galler, P. kupua on Maui, and a leaf galler, P. hualani on Molokai, also suggest these biological shifts happened repeatedly, both within and between islands.
Notable morphological variation is evident for several species, which has resulted in the recognition, in this treatment, of a number of morphological forms. These forms are usually also identifiable in genotypic clusters, but in two widespread species on Oahu, P. oahuensis and P. ohiacola, divergent genetic clusters exist within the same form (Fig. 3). On Hawaii, a widespread species and form, P. pele form pele has two distinct genetic clusters which exhibit strikingly different geographic patterns (P. pele f. pele cluster 1 is composed of regionally distinct clusters, whereas P. pele f. pele cluster 2 is composed of regionally mixed clusters) despite both cluster 1 and cluster 2 occurring across a similar geographic breadth (Fig. 3); a second form, P. pele form kohalensis, is restricted to the Kohala region, and genetic divergence within this form is almost as great as within the much more widespread form pele distributed across the island; greater divergence in form kohalensis could reflect an older and relictual population in Kohala, which is the geologically oldest region of Hawaii. A number of other studies have reported higher genetic diversity in this region, including for the host plant, Metrosideros (Stacy et al. 2014).
Repeated patterns of intraspecific morphological variation in the ohialoha group suggests there may be substantial "standing morphological variation" underlying polymorphism in the genus as a whole. Whether this variation, and the apparent convergences in the ohialoha group, result from a process essentially akin to "morphological drift", Pariaconus. Described species and recognized forms are indicated as well genotypic clusters within forms. Regional locations for sampling sites within islands are shown to give an idea of geographic clustering and isolation (e.g. P. pele form pele cluster 1 has regionally distinct clusters, whereas P. pele form pele cluster 2 is composed of regionally mixed clusters). Support (80-100%) is indicated for major nodes only. or whether selection is involved is not clear. If produced from standing allelic variation (Barrett and Schluter 2008), such polymorphisms may result from rapid adaptation. In the ohialoha group, parallel within species variation in body size, head shape, wing shape, and even genitalic characters, such as size and shape of female terminalia, and shape of male paramere and aedeagus results in some forms from different species appearing superficially similar. In contrast, the "bicoloratus species complex" exhibits something more akin to "morphological stasis" (e.g. P. wyvernus and P. hina), with comparatively little morphological divergence despite similarly high levels of genetic divergence. Crawford (1918) thought that the galling habit in itself indicated an affiliation between Hawaiian and Asian species as there is an unusually high percentage of gallers in some Asian psyllid faunas: in Taiwan and Japan 50% or more of the described psyllid species are gall making, and notably these oriental galling faunas are dominated by taxa in the family Triozidae (Yukawa and Masuda 1996;Yang and Raman 2007;Yang et al. 2013;Percy et al. 2015). What is particularly remarkable in Pariaconus is that within this lineage and over a relatively short period of evolutionary time, there have evolved many different galling habits as well as completely free-living species, and these shifts have all occurred within an insular radiation over a few million years. This makes Pariaconus unique among psyllids.

Origins of Pariaconus and other Myrtaceae-feeders
When considering the age of Pariaconus in the Hawaiian archipelago, Crawford (1918) remarked: "At what time after the establishment of the Ohia lehua [Metrosideros] here the gall psyllids came in is impossible to say, because of the absence of fossils". The kamua group is endemic to the oldest island, Kauai, and recent discovery of subfossils of bicoloratus-type can place these putatively older groups on Kauai with both groups appearing to have diverged early in the evolutionary history of Pariaconus. However, similar degrees of mtDNA divergence within all species groups, as well as the lack of backbone resolution in the genus, implies there was rapid diversification soon after colonization of the Hawaiian Islands. The same range of dates for colonization and diversification by Pariaconus and Metrosideros ) raises the possibility of concurrent in situ progression down the island chain, however this will need to be tested more rigorously with further double-dating analyses (Percy et al. 2004). Different pre-Hawaiian biogeographies: Metrosideros colonizing from the Marquesas (Wright et al. 2001, and Pariaconus apparently colonizing directly from Asia-Australasia via long distance dispersal (Gillespie et al. 2012), precludes a joint colonization event, nevertheless the date ranges suggest Pariaconus arrived in the Hawaiian Islands contemporaneously or soon after Metrosideros.
The closest sister group of the Hawaiian Metrosideros-feeding psyllids remains unknown, but this study confirms a phylogenetic outgroup association with pit-galling species from Australasia and the Palaearctic region on non-Myrtaceae hosts. Rather surprisingly, the pit-galling Metrosideros-feeders from French Polynesia, Fiji, Samoa, and Australasia, as well as other species galling Myrtaceae from Asia and Australasia represent a separate radiation on Myrtaceae. There may yet be an ancestral Myrtaceaefeeding sister taxon to Pariaconus still to be discovered, but it is clear that the Austro-Pacific Myrtaceae-feeders that have been sampled for this study are not close to Pariaconus. Based on the outgroup analysis and character reconstruction, it is an ancestral galling habit that is more likely to have been conserved during the colonization of the Hawaiian Islands, not the association with Myrtaceae. The data presented here strongly support a close relationship with two species in particular, although these are unlikely to be the closest relatives: the Palaearctic pit-galler of oaks, Trioza remota, which has a native distribution range from the UK to Japan (Hodkinson and White 1979, Burckhardt 1989, Önuçar and Ulu 1991, Ossiannilsson 1992, Malumphy et al. 2009), and an undescribed species making pit galls on Banksia in Australia (Tasmania). Trioza remota makes shallow pit galls on the leaves of Quercus; the Australian species makes deep pit galls on the leaves of Banksia, with the immatures situated at the base of the pit, in a similar manner to the cup gallers from Kauai (kamua group). The Banksiafeeder is likely related to two other described Australian species from Banksia (Taylor and Moir 2014), and one of these species, Trioza banksiae Froggatt, 1901, was noted as having free-living immatures (Froggatt 1901). If the Hawaiian lineage is derived from an ancestral lineage that also exhibits within lineage lability in biology, including galling and non-galling taxa, then this ancestral lability could partly explain the propensity for biological shifts in Pariaconus. The pit galling habit and reduced genal processes of these outgroup taxa also support the interpretation of bicoloratus, minutus, and kamua groups as ancestral to the closed galling ohialoha group, with the development of longer genal processes repeatedly derived in Pariaconus.
The question of long distance dispersal to the Hawaiian Islands, whether from another Pacific archipelago or from a continental source, remains an enduring enigma for many insect groups with no obvious means of trans-ocean dispersal. Crawford (1918) envisioned storms as a means of dispersal, stating: "It is conceivable that once in several million years a windstorm might have carried a leaf with galls containing nymphal psyllids and dropped the leaf in an Hawaiian forest of the same kind of trees -an exceedingly rare chance! -whereupon the insect might establish itself". This dispersal scenario seems implausible for immatures as a leaf separated from a plant would soon desiccate resulting in mortality of developing immatures. However, such storms may have delivered adult psyllids to these remote islands; weather system dispersal has been proposed for other organisms, including Metrosideros (Wright et al. 2001, Gillespie et al. 2012.

Does Pariaconus diversification parallel Metrosideros?
Some aspects of Pariaconus diversification suggest parallel and sequential evolutionary progression down the island chain that may mirror patterns found in Metrosideros . One parallel pattern in plant and insects is the basal position of the oldest island of Kauai and higher levels of divergence on this older island versus groups restricted to younger islands. Although there is no evidence for greater molecular divergence on Kauai, there is greater morphological diversity with some taxa more clearly differentiated. Kauai may also have the broadest range of galling habits, though interestingly, the younger island of Hawaii has a similarly high diversity. It should be noted that overall conclusions are limited by the relatively poor sampling on Kauai. Other similarities between diversification patterns of Metrosideros and Pariaconus lineages include the degree of within island ecological and geographic variation (e.g. bog and forest plant morphotypes, and bog and forest psyllid species around Alakai swamp on Kauai); distinct variation found between Waianae and Koolau Mountain ranges on Oahu and between east and west Maui; and the complex patterns, including apparent incipient diversity, on younger islands. These are all suggestive of parallel diversity in plant and insect lineages, which may be an important evolutionary process in this system.
There still remain many questions regarding the underlying mechanisms responsible for morphological variation in Metrosideros polymorpha, despite genetic studies using both nuclear and chloroplast data, and genotype-ecotype analyses , Harbaugh et al. 2009, Wright and Ranker 2010, Stacy et al. 2014. It is likely that the patterns of diversity seen in the psyllid lineage, including species specific preference for one or another host morphotype, are driven in part by the varied phenotypic landscape of the host plant. But whether this effect is unidirectional (i.e. host plant effect on psyllids), or whether the impact and abundance of psyllids could influence shifts in the phenotypic landscape of M. polymorpha (e.g. via frequency dependent selection; Garrido et al. 2016) is an intriguing question. Certainly, the extremely heavy herbivore load on some individual plants caused by one or more Pariaconus species and resulting in tissue necrosis and leaf or bud abscission (Nishida et al. 1980, Gruner et al. 2005 (Fig. 4I), could impact plant fitness and/or reproductive success and thereby influence local ratios of different morphotypes (Bagchi et al. 2014). Plants have evolved a diverse array of anti-herbivory traits, including the evolution of specific defences to specific herbivores (Futuyma and Agrawal 2009), and an increasing number of studies show alternate phenotypes in plants can be driven by interactions with herbivores, such as defensive chemicals or leaf trichomes occurring as a result of herbivory (Agrawal et al. 2012, Hare 2012). In addition, there are a number of studies that have looked at the mechanistic control of alternate phenotypes, for instance via balancing selection and epigenetic functions (Bräutigam et al. 2013, Delph and Kelly 2014, Quintana-Murci 2016. The interplay of plant-insect interactions across the ecotypic-genotypic landscape of the Metrosideros-Pariaconus system could be a new model system that is only just starting to be investigated (e.g. Bailey et al. 2015).

Intra-island divergence in Pariaconus
There are many examples suggesting a strong role for localized isolation by distance in Pariaconus populations, implying limited and localised dispersal within islands. How-ever, it is also possible that migration rates may be higher than observed, but involve a high likelihood of failure to establish within already well established populations (Nosil et al. 2005), which would reinforce population structure and promote local adaptation. High levels of localized population structure on small geographic scales, such as we see in Pariaconus, is not unusual for specialist herbivores versus generalist predators (Rominger et al. 2015), and altitudinal clines in Pariaconus species distributions are partly linked to distributions of preferred host morphotypes (Nishida et al. 1980, Gruner et al. 2005. Even within the same host morphotype, local variation in occurrence and abundance of Pariaconus species is evident where individual Metrosideros plants bear many more galls than neighbouring plants. There are many factors that could influence local insect-plant interactions at this scale (Nishida et al. 1980, Gruner et al. 2005, Stacy et al. 2016, including stochastic processes of ecological turnover, particularly on younger, geologically volatile islands, which may disproportionately impact specialist taxa (Gruner et al. 2007, Hardy andOtto 2014), and contribute to both irregular distributions and rarity. One factor that may explain the mosaic patterns of distribution, at least for galling Pariaconus species, is intraspecific facilitation, whereby gall maker aggregations act as physiological sinks for photosynthate originating elsewhere in the plant (Heard and Buchanan 1998), competition thereafter may promote the utilization of different plant parts (Joy and Crespi 2007) and thus ultimately explain the ecological packing of multiple psyllid species on individual plants. Under this scenario, ecological packing results where tradeoffs are balanced between a) host sharing of favourably modified/induced plant phenotypes, and b) within plant spatial competition. It is possible that the absence of induced resource aggregating effects by the free-living taxa may explain why these species appear to be rarer and more scattered in distribution, although free-living species also frequently co-occur.
Complex patterns of diversification observed on the youngest island of Hawaii (e.g. "bicoloratus species complex") can also be found in a number of other invertebrate groups (e.g. Wessel et al. 2013). One factor that likely plays an important role in promoting species complexes is the instability and fluctuation in the geological landscape on Hawaii, with ongoing volcanic activity and shifting habitats (Goodman et al 2012). What Futuyma (2010) calls "ephemeral divergence", where localized spatial and temporal divergence may be relatively short lived and unstable population structure prevents longer term adaptive signatures becoming established, could be contributing to the disruption of species boundaries and/or maintenance of within species complexes (e.g. P. wyvernus and P. pele). However, similarly complex patterns are observable on most of the older islands that are less volatile geologically (e.g. in P. ohiacola and P. oahuensis on Oahu), thus requiring a broader hypothesis to explain the morphological variability evident within taxa. Crawford (1918) also described notable intraspecific variation in several species, some of which he maintained were incipient species, such as T. lanaiensis, which he considered to be "an incipient species not yet clearly marked from [P. oahuensis]".
A comparison of the patterns of local divergence in Pariaconus with two other Pacific-wide "tramp" species sampled as outgroups: Leptynoptera sulfurea Crawford, 1919 (Triozidae) and Mesohomotoma hibisci (Froggatt, 1901) (Carsidaridae), highlights how different historical processes can generate marked differences in genetic-geographic structure in psyllids. These Pacific-wide psyllid species feed on widespread coastal host plants, Calophyllum inophyllum (Clusiaceae) and Hibiscus tiliaceus (Malvaceae) respectively, and although neither psyllid is native to the Hawaiian Islands, they do have wide native distributions reflecting the wide distribution of their host plants from Asia across the Pacific. The data presented here confirm that specimens of Leptynoptera sulfurea from Taiwan, Singapore, New Caledonia, and French Polynesia have virtually no genetic divergence, whereas specimens of M. hibisci from Singapore, New Caledonia, Norfolk Island, and French Polynesia exhibit considerably greater divergence suggestive of stronger isolation by distance. The geographically wide distributions of both plants are considered native, but anthropogenic influences can not be ruled out, particularly in the case of C. inophyllum, and therefore it is difficult to interpret the differences observed, but these examples nevertheless serve to illustrate that a similar breadth of geographic range may be accompanied by very different genetic structure.

Parallel diversification in Pariaconus
The diversity in Pariaconus is striking considering speciation is not associated with shifts to different plant species, but rather shifts to different biological niches on a single plant species: galling to non-galling, different gall structures and placement of galls on the plant. Crawford (1925) was the first to report rearing adults from galls, and thereby identified species making enclosed galls on leaves as different from those making enclosed galls on the same plants but on stems and buds. Even more striking are characters associated with apparently independent parallel shifts to the same galling habit, and these convergences have led to much taxonomic confusion. An example is the large sized, yellow-green species that gall stems and buds on all major islands (Kauai, Oahu, Maui, and Hawaii). These were previously considered to be the same species or sister taxa. However, in each case where sister taxon relationships are well supported, they are more closely related to a species on the same island with a different galling habit, often a smaller, dark coloured species galling leaves. These shifts within islands appear key to understanding diversification in this system.
An ancestral pit-galling habit may have acted as an evolutionary springboard to both other galling and non-galling habits, with potentially multiple shifts to closed galling or free-living biologies. Independent shifts to closed galling biologies in the kamua and ohialoha groups has resulted in similar changes in immature morphology and chaetotaxy. Among the closed gallers there is evidence of numerous minor parallel evolutionary shifts, and even within species there is some galling lability. In a number of cases, species that predominantly exhibit one galling habit (stem gall, flat leaf gall, cone leaf gall), were found on at least three islands (Kauai, Oahu, Hawaii) to exhibit some galling lability (leaf gallers were galling stems and vice versa). In each of these cases, gall phenotype is faithful to gall position, and the identity of the galler was only detected by DNA sequencing immatures removed from galls. The apparently complex interaction between gall shifts and speciation among these closed gallers reinforces the importance of understanding the "interactome" or "cecidome" (Nabity 2016), in other words the insect-plant-gene interactions and factors influencing plant responses (Nabity et al. 2013, Bailey et al. 2015. In one example on Oahu, the predominantly stem/bud galler (P. oahuensis) produces cone leaf galls in a local population in the northern Koolau Mountains with similar gall structure (except for a different gall opening mechanism) to the cone leaf galler from Hawaii (P. pyramidalis) (compare cone galls in Fig. 50 and Fig. 52), yet these species are not sister taxa; interestingly, on the older island of Oahu it appears to be a more recent and population level shift (from stem galling to galling leaves); whereas on the younger island of Hawaii, the shift is older resulting in a divergent sister taxon from the stem galler (P. hawaiiensis) that produces cone leaf galls (P. pyramidalis). Thus, what likely began with similar local shifts, progressed further on Hawaii to produce two distinct sister taxa. What factors, other than time, promote the evolutionary progression to complete reproductive isolation remains unknown. These examples provide discrete parallel evolutionary systems with which to investigate plant gall shifts. In addition, local lability in galling habit does mean that using gall type alone for identification of taxa (particularly in the ohialoha species group) can be problematic.
As mentioned above, stem/bud gallers have generally larger body size and paler body colour than leaf gallers in the ohialoha group, but there are no other obvious macromorphological traits that are specifically associated with the shifts between leaf and stem galling habits. Variation in ovipositor size between related stem-and leafgalling cecidomyid flies is associated with placement of eggs into plant tissues and deeper implantation on stems (Joy and Crespi 2007), but similarly marked ovipositor differences between pairs of stem-and leaf-galling sister taxa in Pariaconus is unlikely because eggs are deposited on the plant surface, and it is not until 1 st instar feeding commences that gall development, first as a pit gall, is initiated, with completed gall enclosure by the 2 nd instar. This generational progression, from 1st instars in shallow pits before gall enclosure, is also evident on examination of the 1 st instar exuviae, which have typical pit galler morphology (i.e. marginal sectasetae), and can often be found inside closed galls with extant 2 nd instars which no longer retain those traits. It is therefore apparent that galls result in response to psyllid feeding, rather than to oviposition, and preliminary work on the transcriptional landscape of galls using dual psyllid and plant RNA sequence analysis suggests Pariaconus taxa may influence gall development by synthesizing the plant growth hormone auxin (Bailey et al. 2015).
Because gall position is one of the factors determining gall structure, maternal selection of an oviposition site will be important in determining gall type and potentially promoting diversification (Janz et al. 2005, Kato et al. 2010. The cues influencing oviposition site selection may be a critical factor determining rates and direction of diversification in Pariaconus, and oviposition "mistakes" may be an important route to repeated parallel shifts to different plant organs (Joy and Crespi 2007), as well as preference for particular Metrosideros morphotypes (Gruner et al. 2005). Notably, the highly variable egg morphologies remain unexplained, but appear to be relatively phylogenetically conserved, in contrast to homoplasy in overall body size/colour. Further investigation of the more subtle shifts in egg morphology and biology within species (e.g. P. oahuensis) is needed to fully understand the cause and effect of the unusual range of egg types in Pariaconus.

Conclusions
Species diversity in Pariaconus provides a unique example of a psyllid radiation on a single, highly polymorphic host plant. The extraordinary diversity of biologies and morphology found in Pariaconus have emerged within the geological period of the current high islands of the Hawaiian chain, and diversification of psyllid and host plant lineages have occurred within a similar time frame. This raises many questions for future investigation regarding patterns of parallel and convergent evolution, and ecotype-genotype interactions between plant and insect systems over time. Extensive and focused study using a variety of molecular approaches will be needed to explore and understand the complex evolutionary processes in Pariaconus. In this study, the basic patterns of variation in this fascinating group are presented in order to provide a baseline for future investigations.
Adult colour and structure. General body colour either entirely dark (black, brown, or red), entirely or mostly pale (cream, yellow or orange), or distinctly bicolored (pale/ dark) (e.g. Fig. 4R). Overall size variable from ~1.5-4.5 mm in length ( Fig. 4E-G, J-R). Fore wing broadest either in the middle or in the apical third, membrane with or without distinct pattern of pigmentation, if without pattern either clear or fuscous (whiteish opaque in newly emerged adults); veins brown and either with trifurcation of R, M and Cu 1 , or vein R branching slightly anterior (Fig. 4A, this character can vary within populations and even between left and right wings of individuals and is not considered diagnostic for species); vein Rs relatively short, reaching fore wing margin at or proximal to M fork; long to minute setae on fore wing margins and veins; fore wing membrane either with spinules distributed densely throughout all cells or sparsely distributed and limited to a few cells; a cluster of marginal radular spines present in cells cu 1 , m 1 and m 2 ; fore wing apices either acute, bluntly acute, or rounded. Head moderately deflexed downwards, vertex more or less flat dorsally, with lateral ocelli lying on small tubercles, medial epicranial suture distinct; genal processes extremely short to long, and either concolorous, darker, or lighter than general body colour. Antennae short to long; antennal segments 10, with terminal 3(-6) segments usually darker; a single rhinaria apically on segments 4, 6, 8, 9; terminal segment with two unequal length setae. Distal proboscis segment short to long, darker apically. Thorax moderately arched. Minute to long setae on dorsum of vertex and thorax. Legs moderately short and robust to long and slender, tibia longer than femur; hind leg with meracanthus well developed and straight; metafemur with several stout setae apically; metatibia with a cluster of genual spines basally and typically 1+2 (occasionally a single or 1+3) sclerotized apical spurs and a comb of stout unsclerotized setae; tarsi subequal in length ( Fig. 4C-D, H). Male terminalia with more or less rounded subgenital plate; proctiger with moderate or more pronounced posterior lobe medially (interior surface of lobes bearing 4-5 simple setae, Fig. 4B), length shorter, subequal or longer than paramere; paramere variable; distal aedeagus segment apex either hooked or blunt. Female terminalia with proctiger short or long, dorsal surface straight, convex, or medially concave, subequal, shorter or longer than the subgenital plate, long dorsal setae, a simple anal ring composed of a continuous or interrupted double row of cells, and apex acute to bluntly acute; subgenital plate ventral surface either concave or convex, medium to long ventral setae, apex acute to bluntly acute or truncate; ovipositor either with or without serrations. Eggs highly variable, broadly ovoid or slender, pale or dark, with or without microsculpturing and striations, and with or without distinct pedicel and tail. Immatures and biology. Extremely variable immature morphology reflects variation in biologies and galling habits. Immatures may be free-living or gall forming (open and closed galls). Galling species appear to have one immature per gall/gall chamber, but in some galling species dense aggregations of galls result in clusters of chambers in close proximity. Morphologically consistent characters for 5 th instars are antennae with 8-9 segments bearing 4 rhinaria (on segments 4, 6, 8 and 9, or 2 on segment 8) and two short to medium long terminal simple setae, tarsi with broad crescent arolia and each terminal tarsus bearing a long simple or weakly capitate seta, and anus situated ventrally.
Host plant. The host plant of all Pariaconus species is considered to be Metrosideros polymorpha (Myrtaceae). However, there are five described species of Metrosideros in the Hawaiian Islands, and Pariaconus species and galls have occasionally been found on other Metrosideros taxa (e.g. M. macropus, M. rugosa, and M. waialealae) (see Table 2), but in all cases, these Pariaconus species are predominantly on M. polymorpha. Extensive interspecies gene flow among Hawaiian Metrosideros suggests taxonomic concepts in Metrosideros may not reflect discrete genotypic or phenotypic units; the extensive morphotypic variation in M. polymorpha is therefore considered more influential in driving divergence than occurrence on different species of Metrosideros. Nevertheless, more detailed examination of population divergence in Pariaconus species found on multiple Metrosideros species remains to be undertaken.
Comments. Enderlein (1926) erected this genus in order to rectify, as he saw it, the incorrect inclusion of three Hawaiian species (P. nigricapitus, P. minutus, P. gracilis) by Crawford (1918) in the predominantly new world genus Kuwayama based on the absence of genal processes (also referred to as genal cones or genae); Enderlein named this new genus Pariaconus, with the intention of highlighting this homoplasy: 'similarly without cones', yet different from Kuwayama. The name Pariaconus has therefore existed since 1926 as the nomenclaturally correct name for these three Hawaiian taxa, but Pariaconus was not used in subsequent publications (e.g. Zimmerman 1948, Swezey 1954, Nishida et al. 1980, and the use of Kuwayama persisted until noted in a revisionary classification of the Psylloidea by Burckhardt and Ouvrard (2012).
The original placement of some taxa in Kuwayama by Crawford (who also originally described the genus Kuwayama) was done with acknowledged reservations; characters used to define Kuwayama, such as the absence of genal processes (but with swellings below the bases of the antennae), enlarged clypeus, thorax as broad or broader than the head, and wing subacute to acute, are either not found at all, or not consistently found in Pariaconus. The reduced genal processes that are characteristic of the bicoloratus and minutus species groups are usually still visible between the bases of the antennae, and there are no distinct swellings below the antennae. Furthermore, the development of the genal processes in Pariaconus is highly variable and can even vary considerably within species (notably P. gracilis, P. oahuensis, P. ohiacola, and P. pele). Diminutive genal processes are considered the ancestral condition based on data presented here, with development of longer genal processes in more recently derived species (e.g. the ohialoha group).
Pariaconus is a monophyletic genus endemic to the Hawaiian Islands. Four species groups within Pariaconus are recognized: the bicoloratus, minutus, kamua, and ohialoha groups. The taxa are morphologically remarkably diverse, making ancestral outgroup affiliations difficult to interpret, but they are neither allied to the type species of the genus Trioza, nor to Kuwayama. Nor are those members of Pariaconus that were originally assigned to Kuwayama by Crawford (1918) related to other Hawaiian taxa currently placed in Kuwayama that feed on Pisonia (Nyctaginaceae) and Sideroxylon (Sapotaceae) (Caldwell, 1940, Uchida andBeardsley, 1992), nor are they related to other Hawaiian genera feeding on other plant families in the Hawaiian Islands. The Pariaconus species are the only psyllids that feed on the family Myrtaceae in the Hawaiian Islands, but they are not closely related to other Myrtaceae-feeding taxa in the Pacific or Australasia.
The original placement of some of the Pariaconus species in Trioza reflects the use of the genus Trioza as a default placement for triozid taxa with no clear affiliations. The width of the head in Pariaconus is typically greater (bicoloratus and minutus groups) than that of the thorax, or subequal (kamua and ohialoha groups), but the fore wings do not have the long sinuous Rs vein present in the type species of Trioza, T. urticae (Linné, 1758), and Kuwayama, K. medicaginis (Crawford, 1910). The basal metatibial spur arrangement is typically 2+1 in Pariaconus, which is the same as K. medicaginis, but differs from the 3+1 in T. urticae. However, this character, normally considered fixed and di-agnostic of species or even genera, can be variable within populations in Pariaconus (Fig.  4C-D), with one population of P. hawaiiensis including individuals with a single spur, or 2+1, or 3+1. Similar variation was noted by Crawford (1918) for P. oahuensis and P. lanaiensis, see also note on this character in relation to T. eugeniae and T. adventicia.
In this study, broad species concepts are combined with the recognition of morphological forms (infrasubspecific names as per ICZN) to convey the extent and distribution of morphological variation. Some forms recognized here may warrant subsequent recognition at species level.
Adult key to Pariaconus species groups  Fig. 19 (makes pit galls) (on Hawaii) ..... P. minutus (Crawford, 1918) bicoloratus species group A group of mostly diminutive species, though the largest are similar in size to the smaller members of other groups (Fig. 4). This group includes two of the taxa previously placed in Kuwayama (P. nigricapitus, P. gracilis). The immatures develop either in open galls as shallow pits on the leaf surface, or are free-living on the leaf surface. The adults are characterized by reduced genal processes, short antennae, and often distinct bicoloration, with combinations of dark and pale colouration: usually brown/black with yellow/cream, or with individuals entirely dark or entirely pale. Immature morphology is extremely variable. The eggs are typically slender and either with distinct striations and ridges present or entirely lacking, and either with or without a short, laterally positioned pedicel. Currently the greatest diversity of extant species is found on Hawaii, particularly the Kohala region. Subfossils from Kauai and two recently discovered species on Oahu suggest that this group may have been much more diverse on older islands.  (Crawford) in Enderlein (1926): 401.
Adult colour. Typically bicoloured, generally pale cream-yellow to green on thorax and abdomen, head darker, with a dark dorsal stripe from the head extending part or all the length of the body. Fore wing membrane clear or slightly fuscous. Adult structure. Fore wing apex rounded; surface spinules dispersed, usually in all cells but may be reduced or absent in r 1 and c+sc; setae on margins and veins short to minute (Fig. 5A). Antennae short (av. length 0.68; ratio AL:HW av. 1.38); genal processes extremely short (ratio VL:GP av. 4.81); short to minute setae on vertex and thorax; distal proboscis segment short (av. length 0.07); hind tibia subequal to head width (ratio HW:HT av. 1.03) (Fig. 5B, D). Male terminalia (Fig. 5C, H): paramere shorter than proctiger (ratio MP:PL 1.20), broad at base, tapering to narrow neck below apex with short interiorly directed hook; distal aedeagus segment length subequal to paramere (ratio PL:AEL 1.00), base slightly angular and moderately inflated, apex developed into a dorsally flattened, bluntly rounded hook (ratio AEL:AELH 2.00). Female terminalia (Fig. 5E, I-K): proctiger dorsal surface more or less straight, apex blunt, longer than subgenital plate (ratio FP:FSP av. 1.41), anal ring long (ratio FP:RL av. 2.20); subgenital plate with no or slight medial bulge ventrally, apex truncate; ovipositor apex with serrations (3 pronounced lower, and 2 pronounced upper), valvulae dorsalis strongly convex dorsally.
Egg. Unpigmented or light brown, broad and moderately long with uninterrupted striations over entire surface, short pedicel 1/4 length from base, tail moderately long ( Fig. 5F-G).
Immature (note: immature association for P. nigricapitus is based on immatures with the same collection data as the holotype, but given the co-occurrence of taxa, this remains to be confirmed with DNA analysis). Colour and structure 5 th instar: Mid to dark brown. Broadly ovoid in outline with more or less uninterrupted circumference, and the entire dorsal surface raised into a sclerotized dome resulting in a smooth lacquer-like casing ( Fig. 45A-C, E). Fore wing buds with distinct humeral lobes. Tarsi with small reduced claws (Fig. 45D). Circumanal ring broad and shallowly v-shaped, with a single row of elongate cells (Fig. 45A). Four protruding mounds of tissue each terminating in a cluster of distinctly enlarged cells are situated ventro-anterior to the meso and meta coxae ( Fig. 45A; also present in open gallers, e.g. Fig. 48N), similar structures were found in the same position on P. minutus instars within pit galls, and these likely aid either in attachment to the plant surface, or in positional shifting within a gall but are usually no longer visible after preparation for slide mounting and so are illustrated here for the first time using the imprint left in the casing of P. nigricapitus. Chaetotaxy: The entire outer margin is ringed with fused setae resembling reduced sectasetae (Fig. 45A). Dorsum without setae, or with sparse minute simple setae. Host plant notes. Unconfirmed, but may prefer more glabrous morphotypes. Island. Molokai, Lanai, Hawaii. Distribution notes. Although treated broadly as occurring on three islands (based on previous records summarized in Zimmerman 1948), further sampling from Molokai and Lanai is required to confirm this distribution. Only two species from the bicoloratus group were collected on Molokai during this study (P. hina, P. lona). Biology. The immatures described here are considered free-living but develop under a domed casing which remains after the adult has enclosed (see note on need to confirm adult-immature association).
Comments. Belongs to a complex of species within the bicoloratus group ("bicoloratus species complex" Figs 1-2) for which there is scant biological knowledge, but all the taxa may be non-galling. The species complex includes P. nigricapitus, P. hina, and P. wyvernus with notably truncate female genitalia, and three other species (P. nigrilineatus, P. proboscideus, P. kapo). All species appear to have a scattered distribution, occur at low abundance and are infrequently collected; immatures are rarely encountered, and the adult morphology is often cryptic (close examination of slide mounted specimens is required). A complete taxonomic concept of P. nigricapitus remains somewhat uncertain due to the fact that the type material consists of a single entire female, and a partial male specimen with the abdomen missing. Because female morphology is similar among three species in the species complex, the association of the type female specimen with conspecific male counterparts is challenging. Similarly, due to the sympatric co-occurrence of taxa in this species complex, the association of the immature described here, which was collected by O. Swezey in May 1917 with the adult type material of P. nigricapitus, remains to be confirmed.
Type material. Holotype, female (slide mounted, BPBM). See Table 2 for details of type and other material examined for this study.

Pariaconus hina Percy, sp. n.
http://zoobank.org/F97011C2-83EF-419D-9D4E-7C8A337E9CC5 Figures 6, 7 Adult colour. Typically bicoloured, generally pale cream-yellow to green on thorax and abdomen, head darker, with or without a dark dorsal stripe from the head extending part or all the length of the body (Fig. 6K, M, Q). Fore wing membrane slightly to moderately fuscous.
Egg. Unpigmented or light brown, long to moderately long with striations over entire surface, but in form ovostriatus these are more widely spaced and interrupted, with the addition of distinctly raised dorsal ridges, in forms occidentalis and orientalis they are uninterrupted; pedicel absent, tail long in form ovostriatus (and composed of a curious structure of nodules), short in form orientalis, and apparently lacking in form occidentalis (Fig. 7J, L-M).
Immature. Unknown. Host plant notes. Morphotype preference unknown. Island. Molokai, Maui. Distribution notes. Of the three recognized forms, form ovostriatus and form orientalis are currently only known from east Maui, and form occidentalis is only known from west Maui. Only one, diminutive sized female from Molokai was collected, and the distribution and form on this island remain to be confirmed.
Biology. Unknown. Etymology. Named after Hina, a goddess of the moon in Hawaiian mythology (noun in the nominative singular standing in apposition to the generic name).
Comments. Three forms are recognized (Figs 6-7): form ovostriatus (based on the type has a broader paramere, female subgenital plate more truncate and more notably concave apically, and distinct dorsal ridges on the egg), form occidentalis (smallest form, with more slender paramere), and form orientalis (largest from), both the latter have finely striated eggs lacking dorsal ridges. These forms are highly divergent for molecular data (to the extent that there is no support for a monophyletic P. hina) (Fig.  3) but extremely similar morphologically and therefore may represent morphologically cryptic divergence.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.

Pariaconus wyvernus Percy, sp. n.
http://zoobank.org/419BF25F-6552-4CC0-A742-57701EDDB9FC Figure 8 Adult colour. Variable, often strikingly bicoloured with black or dark brown head and pale cream or yellow-green thorax and abdomen, with or without a dark dorsal stripe from the head extending part or all the length of the body, but can also be completely pale throughout (Fig. 8G). Fore wing membrane clear or fuscous.
Egg. (only known for form wyvernus) Unpigmented or light brown, large, with both continuous and interrupted striations over entire surface, apparently lacking pedicel and tail (Fig. 8T).
Immature. Unknown. Host plant notes. Unconfirmed, but may prefer more glabrous morphotypes. Island. Hawaii. Distribution notes. All three forms are found in Kohala, with form wyvernus only known from this region.
Biology. Unknown. Etymology. Named after "wyvern", a mythical winged creature in Medieval mythology, in reference to the rarity and acknowledged taxonomic puzzle this taxon presents (noun in the nominative singular).
Comments. Three forms are recognized (Fig. 8): form wyvernus (based on the type has a longer paramere and larger aedeagus hook), form chimera (larger form has the shortest paramere), and form gorgonus (longer, more slender tibiae, more slender paramere, and bulbous tip to aedeagus hook). The current genetic analyses suggest this taxon may actually be composed of two or more cryptic species that are polyphyletic. Further work is needed, particularly more sampling, to resolve this and therefore recognising this variation with forms is the best option at present.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Immature. Unknown. Host plant notes. Collected from semi-glabrous morphotype growing on lava flow (dated to 1880s) in south western Hawaii.
Island. Hawaii. Distribution notes. Known from only one locality.

Biology.
Unknown, but may be free-living as for P. nigricapitus and P. proboscideus. Etymology. In reference to the dark dorsal stripe that is frequently present in this species and signifies affiliation with the bicoloratus group (adjective in the nominative singular).
Comments. Currently known only from females; females appear particularly similar to P. lona from Molokai, however DNA analysis does not even place these two taxa in the same subgroup of the bicoloratus group, and this exemplifies the need for additional efforts to sample species diversity more completely.
Type material. Holotype female (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Egg. Unknown. Immature. Unconfirmed, but 1 st instars recovered on the surface of leaves at the collection locality have a setal arrangement similar to that illustrated for P. oahuensis (Fig. 50F), with narrow, blunt sectasetae: anterior margin of the head with simple setae only, a single pair of post ocular sectasetae, a single pair of sectasetae on the apices of each wing bud, and the margin of the abdomen with 8 pairs of sectasetae.
Host plant notes. Collected from pubescent morphotypes. Island. Hawaii. Distribution notes. Only known from Kohala. Biology. Unconfirmed, but this species was collected from low growing pubescent forms in upland bog; eggs and 1 st instar immatures were recovered from the plant surface among the trichomes along the mid-rib (upper leaf surface) and petiole, these eggs have widely spaced interrupted surface striations, a short pedicel and a long tail, how- ever, two other bicoloratus species (P. proboscideus, and P. wyvernus form gorgonus) were collected at the same site and therefore association of this egg type remains uncertain.
Etymology. Named after Kapo, a goddess of fertility in Hawaiian mythology (noun in the nominative singular standing in apposition to the generic name).
Comments. Currently known from only one female; this is the largest species in the bicoloratus group and is unusual for the more well developed genae.
Type material. Holotype female (slide mounted, BMNH). See Table 2 for details of type material examined for this study.
Immature. Colour and structure 5 th instar: Appearance is white and spikey (hedgehog-like) due to coverage of stiff white filaments produced from sectasetae (Fig. 45F). Narrowly ovoid in outline with wing buds protruding and distinct humeral lobes . Tarsi with small reduced claws (Fig. 45H). Circumanal ring moderately wide (CPW:RW av. 3.72), and shallowly v-shaped, with a single row of uninterrupted elongate cells (Fig. 45K). Chaetotaxy 5 th instar: Entire dorsal surface and margins covered with pointed sectasetae (Fig. 45G).
Host plant notes. On pubescent and tomentose morphotypes. Biology. This species is free-living on the undersides of pubescent leaves. Etymology. Named for the distinctly longer distal proboscis segment (adjective in the nominative singular).
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Pariaconus poliahu Percy, sp. n. http://zoobank.org/A819F0B0-CAC3-46B5-8FA1-0C7BFC3B124F Figure 12 Adult colour. Typically bicoloured, generally pale yellow to green on thorax and abdomen, head darker black or brown, and a dark dorsal stripe extending part or all the length of the body. Fore wing membrane slightly fuscous.
Adult structure. Fore wing apex rounded; surface spinules distributed in all cells except few or none in r 1 and c+sc; short setae on margins and veins. Antennae short (av. length 0.67; ratio AL:HW av. 1.43); genal processes short and bluntly rounded (ratio VL:GP av. 5.00); short to minute setae on vertex and thorax; distal proboscis segment short (av. length 0.06); hind tibia slender, length subequal to head width (ratio HW:HT av. 1.05). Male terminalia: paramere shorter than proctiger (ratio MP:PL av. 1.22), broad at base and tapering to apex with anteriorly directed hook; distal aedeagus segment longer than paramere (ratio PL:AEL av. 0.75) with base rounded or slightly angular and slightly inflated, and a large, broadly rounded, hooked apex ( Fig. 12H) (ratio AEL:AELH av. 2.39). Female terminalia: proctiger short, dorsal surface convex apically, apex bluntly rounded, anal ring extremely long (ratio FP:RL av.
Egg. Unpigmented, slender, and apparently without striations, pedicel or tail. Immature. Unknown. Host plant notes. Collected from mixed glabrous and pubescent morphotypes. Island. Hawaii. Distribution notes. Only known from the Kohala region of Hawaii. Biology. Unknown. Etymology. Named for Poliahu, a goddess of snow in Hawaiian mythology, in reference to a concept that each snowflake is unique, as many individuals sampled for this species have highly divergent genetic haplotypes (noun in the nominative singular standing in apposition to the generic name).
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Egg. Unpigmented to light brown, short, broad and with striations, mostly uninterrupted, on the dorsal surface, pedicel and tail short (Fig. 13G).
Immature. Unknown. Host plant notes. Collected from glabrous morphotype. Island. Molokai. Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.

Immature. Unknown.
Host plant notes. Collected from glabrous morphotypes. Island. Oahu. Distribution notes. Only known from one locality, the high elevation bog area on Mnt Kaala.
Biology. Unknown. Etymology. Named for Kuini Liliha, a High Chiefess who served the Kingdom of Hawaii as royal governor of Oahu (noun in the nominative singular standing in apposition to the generic name).

Pariaconus gracilis
Adult colour. General body colour typically dark brown to black throughout, but occasionally partly or entirely pale cream or yellow throughout (Fig. 15Q-R). Head often darker than the rest of the body. Individuals are not distinctly bicoloured (e.g. without distinct dorsal stripe). Fore wing membrane clear, or slightly fuscous.
Immature. Colour and structure: Orange or cream. 5 th instar: Narrowly ovoid in outline with wing buds protruding and with distinct humeral lobes (Fig. 46A, D). Tarsi with small reduced claws (Fig. 46D). Circumanal ring shallowly v-shaped, with a single row of elongate cells (Fig. 46C, E). Chaetotaxy: 5 th instar: Margin with medially expanded and overlapping diamond-shaped setae, dorsal surface rugose with ridges but otherwise only minute simple setae (Fig. 46B-C). 1 st instar (Fig. 46F): anterior margin of the head with long simple setae, otherwise marginal setae are narrow, blunt sectasetae (a single pair post ocular, a single pair on the apices of each wing bud, and the margin of the abdomen with approximately 9-10 pairs); by the 2 nd instar, the characteristic diamond-shaped sectasetae are evident around the entire margin.
Host plant notes. Predominantly on pubescent and tomentose morphotypes. Island. Oahu, Molokai, Maui. Distribution notes. A common species on Oahu. Four clusters can be recognized in the DNA analysis: (a) individuals from Molokai and Maui; these in turn group with (b) a population from Oahu's southern Koolau Mnts that have more developed genal processes (Fig. 15G, I). Another cluster includes (c) individuals from both southern and northern Koolau Mnts with moderately reduced genal processes (Fig. 15J, L), and (d) a fourth cluster comprises all individuals from Mnt Kaala, Waianae Mnts, with much reduced genal processes. It appears that this Kaala population is ancestral with increasing development of genal processes being a derived characteristic in Koolau populations; Maui and Molokai specimens have moderately reduced genal processes and appear to be immigrants from the Koolau region of Oahu.
Biology. The immatures are free-living, usually on the lower leaf surface of pubescent and tomentose morphotypes, this host morphotype preference is also noted on slide specimens collected in 1973 by Beardsley (BISH). Eggs appear to be laid mostly singly and sparsely distributed amongst the leaf trichomes.
Comments. One of the most commonly encountered species in the bicoloratus group. Two forms are recognized (Figs 15-16): form gracilis (based on the type is the more common form, with short rounded genae) (Fig. 15J, L), and form conconus (has more developed, apically acute genae) (Fig. 15G, I). Both these forms can be found sympatrically in the Koolau Mountains (Oahu) and form distinct genetic clusters, suggesting some reproductive isolation.
Type material. Holotype female (dry mounted, BPBM). See Table 2 for details of type and other material examined for this study. Variable, usually bicoloured with orange-brown, cream or yellow to greenish-yellow thorax and abdomen, and a dark dorsal stripe from the head extending part or all the length of the body. Fore wing fuscous, especially around anal margin.
Host plant notes. Apparently prefers glabrous morphotypes, with pit galls mostly on the lower leaf surface, occasionally on the upper leaf surface.
Island. Hawaii. Distribution notes. Appears to be widespread; collected from four regions that group into distinct clusters in the DNA analysis: (a) Puu Makaala, (b) Alili plus Kau, (c) Humuula (Hamakua Coast), and (d) the Kohala region (Fig. 3).
Biology. Immatures make pit galls, typically on the lower leaf surface that are initially shallow, becoming deeper with older instars (Fig. 47D-G).
Etymology. Named for the dark dorsal stripe that is frequently present in this species and signifies its affiliation with the bicoloratus group (adjective in the nominative singular).
Comments. One of the largest species in the bicoloratus group. Two forms are recognized (Fig. 17): form kohalensis (based on the type is found in the Kohala region, with broader, shorter paramere, and shorter genae) (Fig. 17J-N), and form communis (more common and widespread, has a more slender, longer paramere, e.g. Fig. 17Q-R [with more round apex in Kau/Alili] and more well developed genae).
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Pariaconus namaka Percy, sp. n. http://zoobank.org/538455FD-5F50-4CA2-810C-63E4A923A424 Figures 18, 47K-U Adult colour. Variable, either entirely pale cream or yellow to greenish-yellow, or with head darker; a partial or weakly marked dorsal stripe extends from the head part or all the length of the body. Fore wing membrane clear or slightly fuscous.
Host plant notes. Collected on glabrous morphotypes. Island. Oahu. Distribution notes. Only known from one locality, the high elevation bog area on Mnt Kaala.
Biology. Immatures make pit galls on the lower leaf surface (Fig. 47R-U), after eclosion the remaining pits are generally shallower than the likely sister taxon on Hawaii, P. dorsostriatus.
Etymology. Named after Namaka, a sea goddess or water spirit in Hawaiian mythology, in reference to the type locality in the wet bog on top of Mnt Kaala (noun in the nominative singular standing in apposition to the generic name).
Comments. A pit-galling habit on Oahu was only recently discovered, previously pit-gallers were only known from younger islands and their presence on Oahu together with subfossils on Kauai (see Discussion) supports an older and more widespread status for the bicoloratus group.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.

minutus species group
Currently this group includes only two species. These are small species, similar to those in the bicoloratus group, but usually entirely one colour, not bicoloured. The eggs are slender and the immatures develop in shallow pits on the upper leaf surface, and occasionally lower leaf surface of pubescent morphotypes of the host plant. This group consistently clusters as sister to the kamua species group rather than together with other pit gallers and free-living species in the bicoloratus group in the molecular analyses.
Adult colour. Variable, typically mid-to dark brown throughout, recently emerged adults can be completely pale cream, head often darker than the rest of the body. A population in the Kilauea Iki caldera (form kilaueaiensis) is typically yellow-orange or dark orange throughout, or occasionally with blue-green abdomens. Fore wing membrane slightly to noticeably fuscous.
Immature. Colour and structure: Smaller immatures are orange and cream, or yellow-brown, larger become blue-green or remain orange (e.g. Kilauea Iki population). 5 th instar: Broadly ovoid in outline and ventro-dorsally flattened with wing buds protruding and distinct humeral lobes (Fig. 46G). Tarsi with small reduced claws (Fig. 46I). Anal ring moderately wide (ratio CPW:RW av. 4.76) and shallowly v-shaped, with a single row of uninterrupted elongate cells (Fig. 46G). Chaetotaxy: 5 th instar: Continuous marginal ring of blunt, weakly bisected sectasetae. Dorsal surface either scattered with small pointed sectasetae (form minutus, Fig. 46G), or small to minute simple setae (form kilaueaiensis). 1 st instar (Fig. 46J): Marginal ring of broad, weakly bisected fan-shaped sectasetae (anterior of head with 11-12 pairs, 1 pair postocular, 1 pair on apices of each wing bud, and 11-12 pairs on abdomen). Apparently only in form kilaueaiensis (Kilauea Iki population) are three pairs of narrow sectasetae present on the dorsal surface of 1 st instars (1 pair on each of head, thorax, abdomen). By the 2 nd instar there is a continuous marginal ring of sectasetae (Fig. 46H).
Host plant notes. Usually on thick leaved, pubescent or semi-pubescent morphotypes. Island. Hawaii. Distribution notes. Widespread on Hawaii. The DNA analysis indicates distinct population clusters, but also more dispersal than for other taxa: two distinct clusters in Kohala, a mixed cluster from Olaa, Kilauea, Kau, and Kona Hema, and another cluster almost entirely from Saddle Road with a single Kau sample.
Biology. Makes pit galls on upper leaf surface. The leaf tissue often forms into a thickened rim of red or yellow around the immature (Fig. 46L-P). The Kilauea Iki population makes somewhat shallower pit galls on the upper leaf surface of low growing shrubs (~1m) in the bottom of the caldera. Eggs are laid individually or in small clusters long the upper leaf margin (Fig. 46K).
Comments. Two forms are recognized (Fig. 19): form minutus (based on the type, with mature adults usually brown, is a smaller form with more slender paramere), and form kilaueaiensis (with adults typically orange, is larger with a shorter, broader paramere).
Type material. Holotype, male (slide mounted, BPBM). See Table 2 for details of type and other material examined for this study.
Egg. Unknown. Immature. Unknown. Host plant notes. Collected from pubescent morphotypes. Island. Maui. Distribution notes. Only known from eastern Maui, in the Makawao area. Biology. Unknown, but likely to be pit galling given the biology of the sister taxon, P. minutus.
Etymology. Named for the more dorsally humped (gibbosus) shape of paramere apex, aedeagus apex, and ovipositor valvulae dorsalis that distinguishes this species from the sister taxon, P. minutus (adjective in the nominative singular).
Comments. Adults of this species are easily confused with P. gracilis in the field. However, Maui is currently the only island where both species occur. Both species are often almost entirely dark brown to black, and similar in overall size.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study. Figure 20. Pariaconus gibbosus sp. n. A fore wing B head C proboscis D head and antenna E hind leg F female G head and thorax H male terminalia I aedeagus and paramere J male terminalia (dorsal view, inward directed apex indicated) K female terminalia L ovipositor (convex shape of valvulae dorsalis indicated).

kamua species group
The kamua species group is a monophyletic group of at least 10 species endemic to the island of Kauai. It includes the largest and some of the smallest species in Pariaconus, as well as the only species in the genus with a distinct fore wing colour pattern (P. melanoneurus), and the only species with a more pronounced lobe on the posterior of the male proctiger, particularly in the largest species in the group (e.g. P. hiiaka). It encompasses perhaps a larger degree of adult morphological variation than found in other species groups, but the immatures are known for only three species, and many of the species are known from only one locality. There is a diversity of galls, including enclosed leaf and stem galls similar to those in the ohialoha group, and a unique type of open gall that is produced on the side of plant stems whereby the inner cambium extrudes out to produce a cup gall. Further study of the different biologies of this species group is required to establish the full range of galling behaviours. Both the endemism of the kamua group on the oldest island of Kauai, and the molecular data, support the interpretation of independent origins of closed galling on Kauai and on the younger islands.
Adult colour. General body colour green, yellow-green or yellow-orange, often with brown on legs, thorax and abdomen. Females may have a darker abdomen due to darkly pigmented egg load. Fore wing membrane clear.
Egg. Mid-to light brown, elongate oval, with longitudinal medial suture entire length of egg (coffee bean-like), surface with microsculpturing and granular in appearance, extremely short pedicel 1/3 length from base, tail lacking (Fig. 21Q).
Immature. Unknown. Host plant notes. Collected predominantly from glabrous and semi-pubescent morphotypes.
Island. Kauai. Distribution notes. Collected in several locations in Kokee State Park, including Alakai, Kalalau and Nu Alolo.
Biology. Based on phylogenetic closeness to P. hiiaka, and the large body size, this species may have a similar closed gall biology, but further study is required to confirm.
Comments. The female lectotype (dry mounted, BMNH) has been examined and compared with the female syntype (dry mounted, BPBM) of Trioza kauaiensis Crawford, 1925. In publications after 1908, the name Trioza iolani has been almost exclusively associated with a common Oahu species (here designated Pariaconus oahuensis). There is no way to avoid this unfortunate synonymization given the following circumstances: Kirkaldy (1902) referred to two specimens in his original description, one from Kauai and one from Oahu, but the two specimens are not from the same species. In the original description, he illustrates only the specimen from Kauai but does not publish a designated type. However, in 1908 he validly lectotypified the Kauai specimen: "The type was from Kauai", and this specimen also bears a hand written label (apparently in Kirkaldy's handwriting), "male type" (though the lectotype is in fact female) (Fig. 21P). The name iolani must therefore be considered to apply to the Kauai specimen, rather than the Oahu specimen, and the Kauai specimen belongs to the same species as that described by Crawford in 1925 as Trioza kauaiensis. Kirkaldy probably had a relatively broad concept of psyllid species (not being a specialist in this group), and he may not have had a good understanding of psyllid morphology because both the specimens and the illustration accompanying the original description (Plate IV, Fig. 2; and shown here in Fig. 21P) are females, not males as Kirkaldy thought them to be. Kirkaldy's other female specimen clearly fits the concept of T. iolani sensu Crawford (1918Crawford ( , 1925 and Zimmerman (1948). Neither Crawford nor Zimmerman appear to have examined the BMNH type material, and Crawford only examined additional material collected by Kirkaldy from Oahu. Nevertheless, Zimmerman (1948) correctly noted that the specimen from Kauai was the type for iolani, and with typical astuteness discerned, "…some confusion regarding the identity of this species. It is generally thought of as one of the commonest psyllids on Oahu, yet the holotype was designated as a Kauai specimen. Further study might reveal that the Oahu form is a distinct species from the Kauai form".
Two forms are recognized on Kauai (Fig. 21): form iolani (based on the type is more common, with paramere apex more extended and posterior shoulder rounded) (Fig. 21H), and form scapulus (with paramere apex short and distinctly extended posterior shoulder) (Fig. 21L). More specimens and an investigation of the biology is required to establish if these are distinct species.
Immature. Colour and structure: Smaller instars orange, larger becoming yellow or blue-green with grey thorax and head. 5 th instar ovoid in outline with wing buds protruding and nondistinct humeral lobes (Fig. 48A, C). Tarsi with large claws (Fig. 48B). Circumanal ring small, u-shape with a single row of often interrupted cells (Fig. 48D). Younger instars are ovate (egg-shaped) with broad head and narrowing abdomen (Fig.  48E). Chaetotaxy: 1 st -5 th instars: Head, thorax and abdomen with scattered short to long simple setae. 1 st instar (Fig. 48E): anterior margin of the head with simple setae, a single pair of short simple setae post ocular, a single pair of short simple setae on the apices of each wing bud, and the margin of the abdomen apparently lacking setae. Host plant notes. Collected predominantly from glabrous and semi-pubescent morphotypes.
Distribution. Kauai. Distribution notes. Collected in two locations in Kokee State Park.
Biology. This species forms enclosed galls on leaves that resemble flat leaf galls in the ohialoha group, but are typically more domed (Fig. 48F-H), and in some cases either convex or concave in the centre (e.g. resembling P. pele form kohalensis, see Fig. 52Z). The galls open by a hinged circular door (Fig. 48I-K) in similar fashion to P. pyramidalis on Hawaii (see Fig. 52W-X). Gall density can severely deform leaves, and cause whole leaf necrosis (Fig. 48F, L). In one location an immature dissected from a stem gall also DNA barcoded to this species (see discussion on galling lability). This species frequently co-occurs with other galling taxa (Fig. 48M).
Etymology. Named after Hiiaka in Hawaiian mythology, the favoured sister of Pele, who dwelled in a sacred Lehua grove and journeyed to Kauai (noun in the nominative singular standing in apposition to the generic name).
Comments. Some variation in paramere shape, particularly in development of anterior shoulder, is illustrated in Fig. 22H-I.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Immature. Unknown. Host plant notes. Morphotype preference unknown, adults collected on both glabrous and pubescent types.

Island.
Kauai. Distribution notes. Known from only one location in Kokee State Park. Biology. Unknown, but its close relationship with P. iolani and P. hiiaka suggest it is likely to make closed galls.
Etymology. Named for the dark pigmentation around the fore wing veins (adjective in the nominative singular).
Comments. This species is the only member of Pariaconus to have a distinctly patterned fore wing. Variation in paramere shape is illustrated in Fig. 23H.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Immature. Unknown. Host plant notes. Morphotype preference unknown, adults collected on both glabrous and pubescent types.
Island. Kauai. Distribution notes. Known from only one location, Kalalau Valley in Kokee State Park. Figure 24. Pariaconus grandis sp. n. A fore wing B fore wing detail C head D male terminalia E paramere F aedeagus and paramere G proboscis H hind leg I female terminalia showing raised anal ring collar (inset) J ovipositor (serrations indicated) K egg (pedicel indicated).

Biology.
Unknown, but morphological affinities with P. hiiaka suggest may make closed galls.
Etymology. The name refers to the large size of the species, it is the largest of the Metrosideros-feeding psyllids in the Hawaiian Islands (adjective in the nominative singular).
Comments. This is the largest of the Metrosideros-feeding species in the Hawaiian Islands, but it is only marginally larger than some of the other large taxa (e.g. P. iolani, P. oahuensis, P. mauiensis, P. hawaiiensis) that are generally yellow-green and probably predominantly stem gallers.
Host plant notes. Found predominantly on glabrous and semi-pubescent morphotypes.
Island. Kauai. Distribution notes. The two recognized forms (brunneis and rubrus) of P. caulicalix are found sympatrically (although form brunneis is more widespread), which, given the molecular differentiation, suggests that, in addition to colour and general size differences that are noticeable in the field, there may be some reproductive isolation.
Biology. This species forms thin-walled cup galls on stems, often clustered together, with one immature per gall chamber (Fig. 49G-J). The gall tissue of the cup is green or yellow-green. Immatures are seated in the base of the cup gall and the ridged sclerotized dorsal surface forms a plug under which is the soft unsclerotized body (Fig. 49G-H).
Etymology. The name refers to the gall type which is a cup (calix) -shaped cambial outgrowth on plant stems (caulae) (adjective in the nominative singular).
Comments. Two forms are recognized (Fig. 25): form brunneis (more common, larger and generally brown, with a longer paramere), and form rubrus (smaller, generally orange-red, and shorter paramere); there are few other morphological characters to distinguish these forms, however there is notable genetic divergence (Fig. 3). This is one of the more common species on Kauai and is closely related to P. crassiorcalix, which makes a thick-walled cup gall on stems of more pubescent morphotypes such as bog ohia in Alakai Swamp. Morphologically, both these species are close to P. lehua (Crawford, 1925).
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.  (Fig. 26G). Fore wing membrane clear or slightly fuscous.
Host plant notes. Only known from one locality where it galls densely pubescent bog morphotypes.
Island. Kauai. Distribution notes. Known only from Alakai Swamp, Kokee State Park. Biology. Forms thick-walled cup galls on stems, galls are often clustered together, with one individual per gall chamber (Fig. 49P-R). The depression forming the cup gall is not as deep as for P. caulicalix, the cup walls are much thicker and the gall tissue is typically dark red or orange-brown. The immature is lodged tightly in the base of the cup with the sclerotized dorsal surface forming a plug under which is the soft unsclerotized body (Fig. 49R). The different structure of the dorsal surface, scaly in P. crassiorcalix and ridged in P. caulicalix, may be related to adaptation to different moisture levels, with P. crassiorcalix found in more humid, wet bog habitat.
Etymology. The name refers to the gall type which is a thick (crassior) -walled and cup (calix) -shaped cambial outgrowth on the plant stems (adjective in the nominative singular).
Comments. This species is a sister taxon to the other known cup gall maker, P. caulicalix, and, together with P. elegans, P. gagneae, P. haumea, and P. lehua, for which biologies are currently unknown, may constitute a sub-clade of cup gallers. A similar type of thick walled cup gall ( Fig. 48N-O, referred to as "a raised button gall on the stem", Russell Messing pers. comm.) is produced by an immature with long waxy dorsal filaments, which may be the immature of one of the described species here, or an as yet undescribed species.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study. (Crawford, 1925), comb. n. Figure 27 Trioza lehua Crawford, 1925: 29 Adult colour. General body colour yellow or orange. Fore wing membrane clear.
Immature. Unknown (see comment under P. crassiorcalix). Host plant notes. Unknown. Island. Kauai Distribution notes. The type location is recorded only as "Nualolo". Biology. The biology of this species is unknown, it may form cup galls on stems as morphologically it is close to the two stem cup-gallers, P. caulicalix and P. crassiorcalix.
Type material. Holotype, male (?) (dry mounted, damaged, abdomen and fore wings missing, BPBM). See Table 2 for details of type and other material examined for this study.
Egg. Unpigmented, short, not sinusoidal, no microsculpturing, pedicel not visible, tail lacking (Fig. 28E) Comments. Known from only one female specimen; the distinctly long, slender terminalia is unlike any other described species.
Type material. Holotype female (slide mounted, BMNH). See Table 2 for details of type material examined for this study. Adult colour. General body colour yellow with darker yellow-brown dorsally. Fore wing membrane clear or slightly fuscous basally.
Egg. Unpigmented to light brown, long and narrow, not sinusoidal, surface with broadly spaced longitudinal striations that are either continuous or interrupted, pedicel appears to be absent, tail lacking ( Fig. 29E-F).
Immature. Unknown.  Etymology. Named after Betsy Gagné to honour her role in promoting biodiversity research, entomology, and conservation in the Hawaiian Islands (noun in the genitive case).
Comments. Known from only one female specimen; the distinctly shaped female terminalia and egg characteristics are not found in other species.
Type material. Holotype female (slide mounted, BMNH). See Table 2 for details of type material examined for this study.
Egg. Unknown. Immature. Unknown. Host plant notes. Collected from glabrous forest tree. Island. Kauai. Distribution notes. Only known location is Alakai, Kokee State Park, in forest near Alakai bog area.
Biology. Unknown. Etymology. Named after Haumea in Hawaiian mythology, the Hawaiian goddess of fertility and mother of Pele (noun in the nominative singular standing in apposition to the generic name).

Comments.
Known from only one male specimen; the distinctly shaped male paramere is not found in other species; morphologically it appears most closely related to P. lehua.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type material examined for this study.

ohialoha species group
The ohialoha species group is a monophyletic clade of species that makes enclosed galls on leaves, stems, and buds. This species group is found on all major islands except Kauai. The group is characterized by typically longer, more acute genal cones (with the exception of P. pyramidalis), longer and usually more slender parameres, and eggs that are distinct for being short, broad, darkly pigmented, lacking surface striations but with microsculpturing, and with pedicels usually long and only slightly off set from base; female abdomens, especially of paler species, can appear much darker when carrying egg loads. Extensive sampling for some of the species in this group reveals that overall body size can be highly variable within species. The immatures are remarkably homogenous with few species specific distinguishing characters. A broad taxonomic approach at species level is combined with an attempt to illustrate the variation within species in this evolutionarily dynamic group.
Due to a high degree of intraspecific variation and inter-island convergence in the ohialoha species group, keys are provided for each island independently (with the exception of Lanai for which insufficient material is available).
Adult key to Pariaconus species in the ohialoha species group found on Oahu Trioza iolani sensu Crawford, 1918Crawford, : 441, 1925 Trioza iolani sensu Zimmerman, 1948: 21 Adult colour. General body colour yellow-green to yellow-brown. Females often appear to have a dark abdomen due to darkly pigmented egg load. Fore wing membrane clear. Adult structure. Fore wing apex rounded; spinules distributed in all cells, but few in r 1 ; medium to long setae on margins and veins (Fig. 31A-D). Antennae long (av. length 1.31; ratio AL:HW av. 2.02); genal processes medium-long (ratio VL:GP av. 1.62), and apically bluntly acute (forms oahuensis and tenuis) or rounded (form latus); medium-long setae on vertex and thorax; distal proboscis segment short (length 0.11); hind tibia longer than or subequal to head width (ratio HW:HT av. 0.90) ( Fig. 31E-P). Male terminalia (Fig. 32A-C): paramere length subequal to or longer than proctiger (ratio MP:PL av. 0.82), broad at the base and more or less parallel-sided (f. oahuensis), tapering (f. tenuis), or inflated medially (f. latus) before constricting to a short neck below an apex with anteriorly directed hook; distal aedeagus segment shorter than paramere (ratio PL:AEL av. 1.18) with base rounded and slightly inflated, and a shallow hooked apex (ratio AEL:AELH av. 2.57). Female terminalia (Fig. 33A-F): proctiger long, dorsal surface more or less straight or slightly convex, apex acute to bluntly acute, anal ring extremely short (ratio FP:RL av. 7.55); subgenital plate with slight medial bulge ventrally, acute to bluntly acute apically; ovipositor apex lacking serrations, valvulae dorsalis moderately or slightly convex dorsally.
Egg. Short, broad, pigmented brown to dark brown (except tip of pedicel and tail) with surface microsculpturing, either with long pedicel and tail, the pedicel with an inflated tip (forms oahuensis and tenuis), in form latus eggs are more slender, lighter in colour with finer surface microsculpturing, a much shorter pedicel without inflated tip, and an unsclerotized patch at the base of the egg likely in the position where the egg contacts the plant surface (in populations making cone leaf galls, the tail is extremely short and there appears to be no pedicel) (Fig. 33G-I).
Immature. Colour and structure: 5 th instar: Cream to orange. Elongate ovoid in outline, wing buds protruding with moderate humeral lobes (Fig. 50A, C). Tarsi with large claws (Fig. 50B). Circumanal ring small (CPW:RW av. 24.40), u-shaped with a single row of sometimes interrupted cells (Fig. 50D). 1 st instars have a scaly dorsal surface (Fig. 50F). Chaetotaxy: 2 nd -5 th instars: Head, thorax and abdomen with scattered long to medium-long simple setae. 1 st instar (Fig. 50F): Setal arrangement similar to those in the bicoloratus group; marginal sectasetae narrow, anterior margin of the head with simple setae, a single pair of post ocular sectasetae, a single pair of sectasetae on the apices of each wing bud, and the margin of the abdomen with 8-10 pairs of sectasetae.
Host plant notes. Primarily on more pubescent morphotypes. Island. Oahu. Distribution notes. The distribution ranges of the three recognized forms overlap, all three are found in both Waianae and Koolau ranges, but rarely at the same collec- tion site suggesting microecological divergence may play a role in explaining this diversity; form oahuensis is the most widespread, form tenuis and form latus are generally less common; however form tenuis appears to be the more common in the southern Waianae and form latus the more common in the southern Koolau, and therefore initial divergence may have taken place allopatrically between the two primary mountain ranges, with subsequent expansion and secondary overlap of ranges.
Biology. Usually galls stems, buds and petioles resulting in irregular swellings (Fig. 50I), but a localized population in the northern Koolau Mnts makes distinct cone galls on leaves resembling those of P. pyramidalis (Hawaii) but with a different opening mechanism (4-5 valves opening from the apex of the gall on the lower leaf surface, versus a circular suture and trap-door opening on the upper leaf surface in P. pyramidalis, compare Fig. 50J-L with Fig. 52T-X). See discussion on lability of galling biology. Dense clusters of gall chambers in bud galls can yield >10 immatures per bud.
Etymology. Named for its distribution on the island of Oahu (noun in the genitive case). Comments. One of the commonest species on Oahu. It can be distinguished most easily from other Oahu species by its typically much larger size and predominantly yellowgreen or yellow-brown colour. This species encompasses Trioza iolani sensu Crawford (1918Crawford ( , 1925 and Zimmerman (1948), see comments under P. iolani (Kirkaldy). Three forms are recognized (Figs 31-33): form oahuensis (based on the type, with variably broad paramere, and shorter, more apically blunt female terminalia), form tenuis (narrower paramere, long slender female terminalia constricted apically), and form latus (with the broad-est paramere that is somewhat medially expanded, long female terminalia not apically constricted, and eggs with unsclerotized base), this latter form apparently also includes the populations that produce cone galls on leaves in the northern Koolau Mnts.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Immature. Colour and structure: 2 nd -5 th instars: Orange or orange-red with cream wing buds. Elongate ovoid in outline, wing buds protruding with moderate humeral lobes (Fig. 50R). Tarsi with large claws. Circumanal ring small (CPW:RW av. 21.78), u-shaped with a single row of often interrupted cells (Fig. 50E). 1 st instar (Fig. 50Q): yellow-brown with scaly dorsal surface. Chaetotaxy: 2 nd -5 th instars: Head, thorax and abdomen with scattered long to medium-long simple setae. 1 st instar (Fig. 50G-H): Margin with broad fan-shaped setae (anterior of head with 5-6 pairs, a single pair post ocular, a single pair on the apices of each wing bud, and 7-8 pairs on the abdomen).
Host plant notes. Mostly associated with glabrous morphotypes. Island. Oahu. Distribution notes. A widespread taxon and probably the most commonly encountered on Oahu, but as with P. oahuensis appears to be undergoing incipient divergence. Forms ohiacola and obtusipterus are the most widespread, found in the Waianae, Aiea, and Koolau mountain regions; form waianaiensis is currently only known from the central Waianae Mnts; form angustipterus is most common in the southern Waianae Mnts, but also occurs in the southern Koolau region.
Biology. Makes flat leaf galls. 1 st instars are found in very shallow pits (Fig. 50Q), usually on the lower surface of young leaves that are often still in bud, the leaf tissue around the instar often becomes red or brown. By the 2 nd instar there is complete enclosure, generating a flat leaf gall type, often with a slight central depression on the lower surface where the original 1 st instar pit was located (Fig. 50O); the 1 st instar exuviae are often found with 2 nd instars in the gall chamber. The scaly sclerotization on the dorsal surface of 1 st instars of the ohialoha group (Fig. 50F) may prevent dehydration during the period when the 1 st instar is on the leaf surface before gall enclosure. Comments. Four forms are recognized : form ohiacola (based on the type has fore wing with apex acute, medium long genae, long paramere, long female terminalia that is apically acute, eggs ovoid with long pedicel), form angustipterus (fore wing more narrow with apex acute), form obtusipterus (largest form, with fore wing apex rounded, shorter genae, long female terminalia), and form waianaiensis (smallest form, with fore wing apex bluntly acute, notably long thin genae, paramere slender, short female terminalia, eggs almost round with short pedicel). In some cases these forms correspond to distinct clusters in the DNA analysis (Fig. 3) and may be good candidates for species recognition with further study of specimens and biology. However, there is substantial variation in fore wing shape, genal process length, and female and male terminalia that are intermediate between these forms, with different combinations found within and between populations making the overall observed pattern complex.
Type material. Holotype, male (dry mounted, BPBM). See Table 2 for details of type and other material examined for this study. (Crawford, 1918), comb. n. Crawford, 1918: 443 Comments. No new material was collected during this study. Below is a summary of the description from Crawford (1918) who considered this species closely related to P. pullatus, yet he also describes it as being "an incipient, not clearly marked, species developing from the Oahuan species [P. oahuensis]"; he suggests that the divergence of P. lanaiensis and P. pullatus may reflect a parallel process to that seen on Oahu between P. oahuensis and P. ohiacola (Crawford 1918). Interestingly, Crawford (1918) reports considerable intraspecific variation in size, noting "it is quite possible that in time these variations will break the species into several distinct ones", and this also reflects the patterns of variation observed on Oahu, Molokai, and Maui.

Trioza lanaiensis
Adult colour and structure. General body colour yellow to brown. Fore wing membrane clear or slightly fuscous, short setae on margins and veins. Reported size is similar to P. molokaiensis, but antennae are reported as up to 3× head width; genal processes long (longer than vertex); male paramere longer than proctiger; female terminalia long (subequal in length to abdomen).
Biology. Unknown, but it likely makes enclosed galls, and if Crawford's hypothesis of parallel divergence to that on Oahu is correct, then this may be a stem galling sister taxon to a leaf galling P. pullatus.
Type material. Holotype, female (dry mounted, BPBM). See Table 2 for details of type material examined for this study. (Crawford, 1918), comb. n. Crawford, 1918: 444 Comments. No new material was collected during this study. Below is a summary of the description from Crawford (1918) who considered this, "an incipient species derived from T. lanaiensis". Additional specimens are needed to test Crawford's hypothesis that it may be a local or seasonal variant of T. laniaensis.

Trioza pullata
Adult colour and structure. Generally body colour dark brown to black, probably the darkest of the ohialoha group. Fore wing membrane clear. Male unknown. Differs from T. lanaiensis in shorter antennae (up to 2× head width), and genal processes (subequal to vertex length), and a shorter Rs vein in fore wing.
Immature. Unknown. Host plant notes. Probably Metrosideros. Original material was collected partly from Cyathodes (Ericaceae) and partly from an undesignated plant.
Island. Lanai. Distribution notes. Known from two localities on Lanai: "Waiopao" ("Waiopaa, west side" in Zimmerman 1948) 29 Nov. 1916, and "undesignated, Dec. 1916and Feb. 1917 Biology. Unknown, but it likely makes enclosed galls, and if Crawford's hypothesis of parallel divergence to that on Oahu is correct (see comment for P. lanaiensis), then this may be a leaf galler.
Egg. Short, broad, pigmented brown to dark brown (except tip of pedicel and tail) with surface microsculpturing, medium-long pedicel with slightly inflated tip, tail short ( Fig. 36Q-R).
Biology. Unknown, but morphology suggests it may gall stems/buds and may have made the bud gall in Fig. 51I.
Comments. Two forms are recognized (Fig. 36): form molokaiensis (based on the type has longer genal processes, more slender paramere, and shorter female terminalia), and form laka (with shorter genal processes, broader paramere, and longer female terminalia). Although there are parallels in characteristics and extent of variation to those observed on Oahu for P. oahuensis, the degree of variation is not as pronounced in P. molokaiensis.
Type material. Holotype, female (dry mounted, BPBM). See Table 2 for details of type and other material examined for this study.

Immature. Unknown.
Host plant notes. Preference unknown; collected from glabrous and pubescent morphotypes.
Island. Molokai. Distribution notes. Known only from Kamakou Preserve. Biology. Unknown; but morphology suggests it may be a leaf galler. Etymology. Named after Hualani, a High Chiefess of Molokai in ancient times (noun in the nominative singular standing in apposition to the generic name).
Comments. Molecular data recovers this taxon as sister to P. kupua from Maui. Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Egg. Short, broad, pigmented brown to dark brown (except tip of pedicel and tail), surface with coarse microsculpturing, and either with distinct medium-long pedicel with slightly inflated tip, or pedicel obscured and an unsclerotized patch at the base of the egg, tail medium-short ( Fig. 38Q-R).
Immature. Unknown. Host plant notes. Collected from glabrous morphotypes. Island. Maui. Distribution notes. Known from east and west Maui; there is some geographical clustering in the molecular data but the east/west divergence of haplotypes is not as distinct as in P. montgomeri. Biology. Unknown; but morphology suggests it may gall stems/buds. Etymology. Named for the distribution on the island of Maui (noun in the genitive case).
Comments. Two forms are recognized (Fig. 38): form mauiensis (based on the type is a smaller form with a shorter, broader paramere, and female terminalia shorter and more apically blunt, known from east Maui), and form kuula (is larger sized, with longer, more slender paramere, and female terminalia longer and apically acute, known from west Maui). There is some variation in the egg type as well; form kuula has an unsclerotized patch at the base of the egg (similar to that found in P. oahuensis form latus) that suggests the forms of P. mauiensis may produce different gall types as in P. oahuensis (e.g. stem galls and cone leaf galls in P. oahuensis).
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Egg. Short, broadly ovoid to almost circular, pigmented brown (except tip of pedicel and tail), surface with microsculpturing, short pedicel with slightly inflated tip, tail extremely short.
Host plant notes. Collected from glabrous morphotypes. Island. Maui. Distribution notes. Known from east and west Maui; the molecular analysis clearly distinguishes eastern from western haplotypes.
Biology. Galls stems and occasionally petioles, resulting in irregular swellings.
Etymology. Named after the kupua, tricksters of the island forests in Hawaiian mythology (noun in the nominative singular standing in apposition to the generic name).
Comments. Molecular data recovers this taxon as sister to P. hualani from Molokai. Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study.
Adult structure. Fore wing apex bluntly acute to rounded; spinules distributed in all cells but few in r 1 ; medium-short setae on margins and veins. Antennae medium-long (av. length 1.06; ratio AL:HW av. 1.89); genal processes medium-short (ratio VL:GP av. 2.38), and bluntly acute; medium-short to short setae on vertex and thorax; distal proboscis segment short (av. length 0.08); hind tibia length subequal to head width (ratio HW:HT av. 1.03). Male terminalia: paramere length subequal to or longer than proctiger (ratio MP:PL av. 0.95), slender, broader at the base and more or less parallel-sided or slightly medially expanded before constricting below apex with a small anteriorly directed hook; distal aedeagus segment shorter or subequal to paramere (ratio PL:AEL av. 1.05) with base rounded, not or slightly inflated, and a short, compact, shallow hooked apex ( Fig. 40J-K) (ratio AEL:AELH av. 2.67). Female terminalia: proctiger long, dorsal surface slightly to moderately convex, apex acute, anal ring short (ratio FP:RL av. 5.24); subgenital plate with slight to moderate medial bulge ventrally, acute apically; ovipositor apex lacking serrations, valvulae dorsalis moderately convex dorsally.
Egg. Short, broad, pigmented brown to dark brown (except tip of pedicel and tail) with fine surface microsculpturing, medium-long pedicel with slightly inflated tip, tail extremely short or absent (Fig. 40O).
Host plant notes. Collected from glabrous morphotypes. Island. Maui. Distribution notes. Known from east and west Maui; molecular data indicates very distinct eastern and western populations that are characterized by the two recognized forms. Biology. Makes flat leaf galls (Fig. 51G-H), often with a small central depression (see also P. ohiacola).
Etymology. Named after Steve Montgomery, an extraordinary field biologist who made a substantial contribution to this study (noun in the genitive case).
Comments. Two forms are recognized (Fig. 40): form montgomeri (based on the type is a slightly larger form, with more sinuous paramere, less developed aedeagus hook, and longer female terminalia, known from west Maui), and form paliuliensis (generally smaller with paramere more straight, aedeagus hook more developed, and shorter female terminalia, known from east Maui).
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study. (Crawford, 1918), comb. n.  Crawford, 1918: 444 Adult colour. General body colour green or yellow-green, but often with brown on legs, thorax and abdomen. Females often appear to have a dark abdomen due to darkly pigmented egg load. Fore wing membrane clear.
Egg. Short, broad, pigmented brown to dark brown (except tip of pedicel and tail), surface with microsculpturing, medium-long pedicel with slightly inflated tip, tail short (Fig. 41I-K).
Host plant notes. Known from both glabrous and pubescent morphotypes, but mostly associated with pubescent and semi-pubescent types.
Island. Hawaii. Distribution notes. Widely distributed on the island of Hawaii; there are some distinct clusters in the molecular analysis, with the southern and western populations sister to and nested within members from Kohala (north), and these are distinct from two individuals from the central Saddle Rd area.
Biology. Galls stems and buds, resulting in irregular swellings, and in the case of buds, inhibits bud growth (Fig. 52O-R).
Comments. Individuals from Kona Hema have a notably broader paramere. Type material. Holotype, male (?) (dry mounted, damaged, abdomen missing, BPBM). See Table 2 for details of type and other material examined for this study.
Egg. Short, broad (almost circular in Kohala form), pigmented brown to dark brown (except tip of pedicel and tail) with surface microsculpturing, medium-long pedicel with slightly inflated tip, tail extremely short or absent (Fig. 43H-I).
Immature. Colour and structure: 5 th instar: Cream to orange. Elongate ovoid in outline, wing buds protruding with moderate humeral lobes (Fig. 52K). Tarsi with large claws. Circumanal ring small (CPW:RW av. 23.67), u-shaped with a mostly single row of sometimes interrupted cells (Fig. 52K-L). 1 st instar dorsal surface scaly. Chaetotaxy: 5 th instar: Head, thorax and abdomen with scattered long to medium-long simple setae, sometimes with clusters of stouter simple setae on the abdomen (Fig. 52K, M). 1 st instar (Fig. 52N): Setal arrangement similar to P. ohiacola but with marginal sectasetae narrow and blunt as in P. oahuensis (anterior of head with 4-5 pairs, a single pair post ocular, a single pair of sectasetae on the apices of each wing bud, and the abdomen with 7-8 pairs).
Host plant notes. Known from both glabrous and pubescent morphotypes, but mostly associated with glabrous and semi-pubescent forms.
Island. Hawaii. Distribution notes. The most common species on Hawaii. In addition to a distinct form "Kohala" only known from the Kohala region, within the "common" form there are two groups both of which are widespread: group 1 includes distinct populations from Kona Hema (south west), Kohala (north east), Saddle Rd (central), and Hualalai (north west); group 2 is more mixed with individuals from south, central, east and west, but not from Kohala. Biology. Makes flat galls on leaves (Fig. 52X-BB), galls often have a small depression or raised plug in the centre (Fig. 52Z-AA). Galls open typically on the lower leaf surface either by irregular cracking, or more rarely by a circular suture around the margin of the gall (Fig. 52Y-BB). This species often co-occurs with P. hawaiiensis (e.g. Fig. 4I) and/or P. pyramidalis (e.g. Fig. 52X-Y).
Etymology. Named after Pele, the volcano and fire goddess in Hawaiian mythology (noun in the nominative singular standing in apposition to the generic name).
Comments. Two forms are recognized : form pele (based on the type, is the most common form, generally smaller with slightly shorter genae, longer paramere and longer female terminalia, the egg is more ovoid), and form kohalensis (slightly longer genae, shorter paramere and shorter female terminalia, the egg is almost round and the pedicel shorter). Comparison of paramere shape and size is illustrated in Fig. 43G.
Type material. Holotype male (slide mounted, BMNH). See Table 2 for details of type and other material examined for this study. Adult colour. General body colour brown or yellow-brown, or yellow-green, the genal processes are often paler than the head. Females often appear to have a dark abdomen due to darkly pigmented egg load. Fore wing membrane clear, or slightly fuscous.
Egg. Short, broad, pigmented brown to dark brown (except tip of pedicel and tail) with surface microsculpturing, long pedicel with inflated tip, tail long (Fig. 44I).
Immature. Colour and structure: 5 th instar: Cream to orange. Elongate ovoid in outline, wing buds protruding with moderate humeral lobes (similar to P. hawaiiensis in Fig. 52A). Tarsi with large claws. Circumanal ring small (CPW:RW av. 16.96), u-shaped with patches of single or multiple rows of interrupted cells (Fig. 52D-E), sometimes reduced or absent. Chaetotaxy: 5 th instar: Head, thorax and abdomen with scattered long to medium-long simple setae (Fig. 52F, I). 1 st instar (Fig. 52H): Setal arrangement similar to P. oahuensis, with simple setae on anterior margin of head and otherwise narrow, blunt sectasetae (a single pair post ocular, a single pair on the apices of each wing bud, and 7-8 pairs on the abdomen); by the 2 nd instar all setae are simple (as is typical of ohialoha group) (Fig. 52G).
Host plant notes. Known from both glabrous and pubescent morphotypes.         cone galls on glabrous host morphotypes clustered along the leaf mid-vein T detail of narrow cone gall produced on glabrous host morphotype U, V broad cone galls produced on more pubescent host morphotypes W scars remaining on upper leaf surface from old cone galls X-Y single leaf with both P. pele and P. pyramidalis galls: X upper leaf surface with cone galls opening by hinged circular door (indicated) Y lower leaf surface with flat galls opening by valves (indicated) Z donut-type gall with central depression produced by P. pele form kohalensis AA-BB P. pele gall variations: AA gall with central plug (indicated), opening with circular suture BB gall produced on the leaf margin.