Cladistic analysis, phylogeny and biogeography of the Hawaiian Platynini (Coleoptera: Carabidae

James Liebherr

The 128 known native Hawaiian species of the tribe Platynini are analysed cladistically. Cladistic analysis is based on 206 unit-coded morphological characters, and also includes forty-one outgroup taxa from around the Pacific Rim. Strict consensus of the multiple equally parsimonious cladograms supports the monophyly of the entire species swarm. The closest outgroup appears to be the south-east Asian-Pacific genus Lorostema Motschulsky, whose species are distributed from India and Sri Lanka to Tahiti, supporting derivation of the Hawaiian platynines from a source in the western or south-western Pacific. The biogeographic relationships of the Hawaiian taxa are analysed using tree mapping, wherein items of error are minimized. The area cladogram found to be most congruent with the phylogenetic relationships, and most defensible based on underlying character data is {Kauai[Oahu(Hawaii{Lanai[East Maui(West Maui + Molokai)]})]}. This progressive vicariant pattern incorporates progressive colonization from Kauai, and vicariance of the former Maui Nui into the present islands of Molokai, Lanai, West Maui and East Maui. The evolution of flightlessness, tarsal structure, pronotal setation and bursal asymmetry are evaluated in the context of the cladogram. Brachyptery is a derived condition for which reversal is not mandated by the cladogram, although repeated evolution of reduced flight wings is required. Tarsal structure supports Sharp's (1903) recognition of Division 1 as a monophyletic assemblage, but exposes his Division 2 as a paraphyletic group requiring removal of the genus Colpocaccus Sharp. Pronotal setation is exceedingly homoplastic, and is not useful for delimiting natural groups. Left-right asymmetry of the bursa copulatrix reversed twice independently, resulting in mirror-image bursal configurations in B. rupicola and Prodisenochus terebratus of East Maui. The amount of character divergence is greater among species comprising Division 1 than among species of its sister group, the redefined Division 2. Based on superior fit of Division 1 relationships to the general biogeographic pattern, a greater speciation rate coupled with more extensive extinction is rejected as the cause for this greater divergence. Intrinsic differentiation in the processes underlying cuticular evolution appears to be more consistent with the observed biogeographic and morphological patterns.

Systematic Entomology (1998) 23, 137–172 Cladistic analysis, phylogeny and biogeography of the Hawaiian Platynini (Coleoptera: Carabidae) J A M E S K . L I E B H E R R and E L W O O D C . Z I M M E R M A N * Department of Entomology, Cornell University, Ithaca, New York, U.S.A. and *Division of Entomology, C.S.I.R.O., Canberra, A.C.T., Australia Abstract. The 128 known native Hawaiian species of the tribe Platynini are analysed cladistically. Cladistic analysis is based on 206 unit-coded morphological characters, and also includes forty-one outgroup taxa from around the Pacific Rim. Strict consensus of the multiple equally parsimonious cladograms supports the monophyly of the entire species swarm. The closest outgroup appears to be the south-east Asian-Pacific genus Lorostema Motschulsky, whose species are distributed from India and Sri Lanka to Tahiti, supporting derivation of the Hawaiian platynines from a source in the western or south-western Pacific. The biogeographic relationships of the Hawaiian taxa are analysed using tree mapping, wherein items of error are minimized. The area cladogram found to be most congruent with the phylogenetic relationships, and most defensible based on underlying character data is {Kauai[Oahu(Hawaii{Lanai[East Maui(West Maui 1 Molokai)]})]}. This progressive vicariant pattern incorporates progressive colonization from Kauai, and vicariance of the former Maui Nui into the present islands of Molokai, Lanai, West Maui and East Maui. The evolution of flightlessness, tarsal structure, pronotal setation and bursal asymmetry are evaluated in the context of the cladogram. Brachyptery is a derived condition for which reversal is not mandated by the cladogram, although repeated evolution of reduced flight wings is required. Tarsal structure supports Sharp’s (1903) recognition of Division 1 as a monophyletic assemblage, but exposes his Division 2 as a paraphyletic group requiring removal of the genus Colpocaccus Sharp. Pronotal setation is exceedingly homoplastic, and is not useful for delimiting natural groups. Left-right asymmetry of the bursa copulatrix reversed twice independently, resulting in mirror-image bursal configurations in B. rupicola and Prodisenochus terebratus of East Maui. The amount of character divergence is greater among species comprising Division 1 than among species of its sister group, the redefined Division 2. Based on superior fit of Division 1 relationships to the general biogeographic pattern, a greater speciation rate coupled with more extensive extinction is rejected as the cause for this greater divergence. Intrinsic differentiation in the processes underlying cuticular evolution appears to be more consistent with the observed biogeographic and morphological patterns. Introduction David Sharp’s (1903) revision of the Hawaiian Platynini was a culmination of substantial 19th century activity in the group. Initial descriptive efforts by the Rev. Thomas Blackburn (1877, 1878, 1879, 1881) combined names for native Hawaiian Correspondence: Dr James K. Liebherr, Department of Entomology, Comstock Hall, Cornell University, Ithaca, NY 14853–0901, U.S.A. E-mail: jkl5@cornell.edu © 1998 Blackwell Science Ltd platynines with the then recognized European genus name Anchomenus Bonelli. Nonetheless, the vast divergence in body form observable among the Hawaiian platynine species (Figs 1– 6) also led to descriptions of endemic genera by both Blackburn (1878, 1884) and Sharp (1878, 1884). By the time of Sharp’s (1903) treatment of the group in the Fauna Hawaiiensis, extensive field surveys by R. C. L. Perkins (Perkins, 1913; Manning, 1986; Liebherr & Polhemus, 1997) had added substantially to the number of taxa discovered by Blackburn, leading Sharp to divide the fauna into twenty-three genera. Tarsal configuration was the basis for fundamental division of 137 138 James K. Liebherr and Elwood C. Zimmerman Figs 1–6. Native Hawaiian platynine Carabidae: 1, Apteromesus maculatus; 2, Brosconymus optatus; 3, Atrachycnemis perkinsi; 4, Deropristus blaptoides; 5, Mesothriscus opacus; 6, Atelothrus cheloniceps. Sharp’s Anchomenides; his Division 1 comprised species with dorsally convex tarsal articles (Figs 1–4, 78, 80), his Division 2 comprised species with two longitudinal laterodorsal sulci on the basal tarsomeres (Figs 5, 6, 76, 77). Genera within divisions were diagnosed chiefly by flight wing configuration and pronotal chaetotaxy. He admitted that the use of setation might lead to an artificial classification, for ‘an individual, by a simple process of discontinuous variation, such as there is reason to believe actually occurs, may ipso facto pass from the genus of its parents to another’ (Sharp, 1903: 176– 177). Nonetheless, he felt that study of such ‘dislocations of taxonomy’ might provide ‘some evidence of real importance as to the mode of origination of species and of genera’ (Sharp, 1903: 177). Sharp believed the Hawaiian platynine fauna to be derived from multiple colonizations: ‘I do not believe in one importation as the origin of the Hawaiian Anchomenidae. I think there have been at least four, 1 I know no reason, except your impression, to which I attach much importance, why there should not have been more’ (Sharp, 1900). Finally, he wrote that ‘The Kauai species of Anchomenides are as a rule remarkably distinct; and it is not unreasonable to infer that ... it has been comparatively free from immigration from the other islands, though not infrequently sending emigrants to them’ (Sharp, 1903: 189). In this contribution, we report a cladistic analysis of the currently known 128 species of native Hawaiian Platynini. The © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini phylogenetic relationships implied by the cladistic analysis indicate that the Hawaiian platynine fauna (1) is monophyletic, (2) has its earliest known diverging lineages on the oldest high island of Kauai and (3) has radiated so that endemic species occupy all of the present high islands of Kauai, Oahu, Molokai, Lanai, West and East Maui and Hawaii. As nearly all of the species are precinctive to a single island, the phylogenetic hierarchy of the species serves well as the basis for deducing the group’s biogeographic history. We utilize the tree mapping method implemented by Page (1990, 1993) to explore the basis for the biogeographic patterns exhibited by the Hawaiian platynines. Given its extreme isolation, great topographic diversity and geographical position among moist tropical trade winds, the archipelago of Hawaii hosts a wide array of speciose radiations (Zimmerman, 1948; Carlquist, 1965; Wagner et al., 1990). Windward mesic montane forests and wet rainforests contrast strongly with drier leeward scrub forests, providing a wide variety of habitats supporting over 900 native plant species and many associated animals. The extreme isolation of Hawaii strongly supports the supposition that many of these speciose radiations are monophyletic. However, demonstrating such monophyly has not generally proved easy. For arguably the best studied radiation – drosophilid fruit flies (Diptera: Drosophilidae) – substantial differences exist among hypotheses of relationships for the major lineages on Hawaii: the genus Scaptomyza Hardy, and the Hawaiian ‘Drosophila’, or Idiomyia Grimshaw sensu Grimaldi (1990). Morphological data support independent origins for these radiations (Grimaldi, 1990; DeSalle & Grimaldi, 1993). Mounting molecularly based data suggest otherwise that the entire array of native Hawaiian fruit flies is monophyletic (DeSalle & Grimaldi, 1991; DeSalle, 1995). The monophyly of Megalagrion MacLachlan damselflies (Odonata: Coenagrionidae) was assumed prior to cladistic analysis based on larval and adult morphological characters (Polhemus, 1997). Other speciose radiations, such as the plagithmysine long-horned beetles (Coleoptera: Cerambycidae), have also been assumed to be derivatives of a single mainland ancestor (Gressitt, 1978). Conversely, the spider genus Tetragnatha Walckenaer (Araneae: Tetragnathidae) comprises a large number of species on the islands that may have arisen from at least three separate founder events (Gillespie et al., 1994). Asquith (1994), using another approach for the metrargine Lygaeidae (Hemiptera), discounted morphological data that suggest polyphyly of the tribe in Hawaii, and accepted a less parsimonious phylogenetic hypothesis that supported a single origin for the fauna. From these few examples, it is apparent that characters unambiguously diagnosing monophyletic Hawaiian radiations are not overly abundant (e.g. Kaneshiro et al., 1995). Our analysis suggests a reason for this situation: character evolution at the origin of an explosive radiation does not differ from that occurring throughout the radiation. Homoplastic changes in characters occurring in near outgroups are repeated during differentiation of speciose radiations. Therefore, no single character or suite of characters can unambiguously diagnose all members of Hawaiian radiations. Only a comprehensive analysis utilizing information from all characters, analysed © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 139 using logically consistent methodology, can permit unambiguous demonstration or refutation of monophyly for such speciose radiations. This study undertakes such a mission. The current study of Hawaiian platynines was conducted in the context of platynine taxa from around the Pacific Ocean. The platynine fauna of Southeast Asia and the Australian region (Wallacea) figures prominently as the source for the Hawaiian radiation. However, at present we know almost nothing about the phylogenetic relationships of Australasian platynine lineages. Darlington’s (1952, 1971) treatment of the New Guinea Platynini comprises the most comprehensive analysis of distinct lineages among the Wallacean platynine fauna. Subsequent revisionary work, such as Louwerens (1953), has grouped many lineages under the generic name Colpodes MacLeay, although it is now clear that this genus name, as currently applied, artificially groups a non-monophyletic assemblage of taxa (Liebherr, 1998). The extensive inclusion of Oriental, Australian and Pacific outgroups in this analysis serves as a first attempt to delineate natural groups among them, as well as providing the context for determining the origin of the Hawaiian platynine fauna. Materials and methods Taxa The native Hawaiian Platynini comprises 128 known specieslevel taxa (unpublished data), all of which were included in this cladistic analysis. One taxon, Metromenus fraternus, was represented by two terms, the nominate form and a form considered to be Metromenus aequalis by Sharp (1903), but currently considered to be a conspecific pale colour variant of M. fraternus (unpublished data). Forty-one outgroup taxa were included, representing platynine taxa from around the Pacific Rim. For all analysed species, the names, authors and geographical provenance are provided in Figs 86–88. For both ingroup and outgroup taxa, undescribed species are included in the analysis, and are designated by a provisional species name in parentheses, preceded by the abbreviation ‘nsp.’. These names are not to be construed as valid combinations nor any form of uninomial species epithet. For all ingroup Hawaiian taxa, we are preparing a separate revision including keys, diagnoses, descriptions, new combinations, replacement names and synonymies, all based on recognized Linnean nomenclature (Liebherr & Zimmerman, unpublished data). The Fijian undescribed taxa, nsp. ‘apterum’, nsp. ‘peckorum’, and nsp. ‘opacidermis’, are to be described by B. P. Moore (personal communication). The undescribed species from Vanuatu, nsp. ‘erythroderus’ and nsp. ‘multipunctatus’, represent manuscript names of F. I. van Emden on material in the Museum National d’Histoire Naturelle, which the senior author will describe. The undescribed species nsp. ‘agonoides’ was first included as an ingroup taxon, as the only two specimens are labelled, HI: Haleakala W edge: above Kolua, 2400 m, J. L. Gressitt (BPBM). Its character distribution argues against its membership in the Hawaiian platynine radiation, meaning that the specimens have been mislabelled. Nonetheless, it was 140 James K. Liebherr and Elwood C. Zimmerman © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini retained as an outgroup to assist in elucidating its true provenance. Seven of the 128 native Hawaiian species were known only from males, and six were known only from females. One species, known only from a teneral male specimen, nsp. ‘kipahulu’, was not dissected nor scored for male aedeagal characters. Two of the forty-one outgroup species were known only from males, and four only from females. Colpodes brunneus, the type species of Colpodes, is known only from the single female type; this specimen was not dissected. Characters The cladistic analysis was based on 206 unit-coded morphological characters (Farris et al., 1970), forty-five from the female reproductive tract and ovipositors, twenty-three from the male aedeagus, and 138 from external cuticular structures (Appendix 1). Unit characters are grouped in sets that define the various configurations of observed structures. This method was used so that we could group character states into strictly hierarchical series, i.e. additive coding; or combinations of hierarchical and unordered states, i.e. additive or non-additive coding (Farris et al., 1970). For multistate characters, only those character-state combinations observed in one or more taxa were used, thereby ensuring that internal nodes of the cladogram were optimized to observed characterstate combinations (Farris, 1970). In several cases (characters 91–96, 105–110 and 169–172), we relaxed a standard assumption of numerical cladistics (Farris et al., 1970) by permitting reticulate character state trees. Such coding admits that homoplasy must exist to account for the observed states, but we felt that the pattern of homoplasy should be determined by a simultaneous analysis, using phylogenetic information from all characters. One autapomorphy, character 148, is included. Its inclusion is explained under characters 146–150. Character 48 is autapomorphous in one taxon, but due to the lack of males for several potential sister taxa, it was included to predict male states for those taxa based on information from the other characters. Character 168 is autapomorphously present in one taxa and polymorphic in a second. All other characters have at least two taxa exhibiting either state. Characters missing 141 from taxa for reasons such as lack of specimens of a particular sex, or characters that were polymorphic for a terminal taxon are considered ambiguous (?) in the matrix. Inapplicable characters are indicated with a dash (–) in the matrix. When characters were based on division of continuous measurements, or were grouped into complex transformation series, we have added explanations below. The establishment of homology and discussion of patterns of character evolution is dependent upon the resultant cladogram, and these issues are addressed in the Discussion. Initial character polarity, as reported in the character descriptions, was designated relative to a platynine groundplan. Rooting the cladogram at the outgroup established polarity for all characters on the cladogram (Nixon & Carpenter, 1993). Dissection, staining and microscopical methods are described in Liebherr (1992). Measurements of female reproductive tract characters were done with an ocular micrometer in a phasecontrast compound microscope. Female reproductive tract characters 0–1. Apical gonocoxite with one dorsal ensiform seta (0,0) (Figs 23–25, 27, 31–33); with two dorsal ensiform setae, at least unilaterally (1,0) (Figs 26, 28–30); lacking dorsal ensiform setae (0,1). 2–4. Lateral ensiform setae of apical gonocoxite moderately elongate, ù 0.123 to ø 0.223 length apical gonocoxite (0,0,0) (Figs 26, 29–32); lateral ensiform setae peglike, ù 0.103 to , 0.123 length gonocoxite (1,0,0) (Figs 23, 25, 27); lateral ensiform setae very small, ø 0.083 length gonocoxite (1,1,0) (Figs 24, 28); lateral ensiform setae very large, broad, . 0.233 length gonocoxite (0,0,1) (Fig. 33). The shape of the lateral ensiform setae, from very small, to peglike, to moderately elongate, to large and broad, is reflected in their increasing length relative to the overall length of the apical gonocoxite. The longest of the ensiform setae was measured from its median base, sometimes situated mesad to the lateral margin of the apical gonocoxite (e.g. Fig. 27), to its apex. The apical gonocoxite length was measured as the straight-line distance from the mediobasal articulatory condyle to the tip. 5. Apical gonocoxite tip acuminate (0) (Figs 23, 31); tip rounded (1) (Figs 28–30). Figs 7–33. Female reproductive structures of Hawaiian platynines: 7, hypothetical mosaic reproductive tract, ventral view, including spermatheca (sp) of Apteromesus maculatus, apical bursal pouch (ap) of Atrachycnemis perkinsi, and bursal glands (bg) of Chalcomenus costatus (lt 5 laterotergite of abdominal segment IX, sd 5 spermathecal gland duct); 8, spermatheca, ventral view, Colpocaccus posticatus; 9, spermatheca, ventral view, Brosconymus optatus; 10, spermatheca, ventral view, Disenochus aterrimus; 11, spikelike lumenal bursal spicules, e.g. Metromenus sphodriformis; 12, triplet cristate lumenal bursal spicules, Mauna frigida; 13, pentuplet bursal spicules, e.g. Disenochus fractus; 14, heptuplet bursal spicules, e.g. Atelothrus aaae; 15, rounded, sclerotized dorsal bursal pouch, Chalcomenus corruscus; 16, rounded dorsal bursal pouch, Chalcomenus costatus; 17, quadrate dorsal bursal pouch, Broscomimus lentus; 18, narrow, keyhole-shaped dorsal bursal pouch, Brosconymus optatus; 19, collarlike dorsal bursal pouch, Metromenus audax; 20, irregularly folded dorsal bursal pouch, dorsal view, nsp. ‘mandibularis’; 21, basolaterally winged dorsal bursal pouch, Metromenus caliginosus; 22, basolaterally and apicolaterally winged dorsal bursal pouch, Atelothrus longulus; 23, right gonocoxa, with basal gonocoxite (gc1), and apical gonocoxite (gc2) bearing two lateral and one dorsal ensiform setae, and two apical nematiform setae, ventral view, Disenochus aterrimus; 24, right gonocoxa, ventral view, Apteromesus maculatus; 25, right gonocoxa, ventral view, nsp. ‘bryophila’; 26, right gonocoxa, ventral view, Disenochus sulcipennis; 27, right gonocoxa, ventral view, nsp. ‘hihia’; 28, right gonocoxa, ventral view, Barypristus incendiarius; 29, right gonocoxa, ventral view, Atrachycnemis sharpi; 30, right gonocoxa, ventral view, Deropristus blaptoides; 31, left gonocoxa, ventral view, Colpocaccus posticatus; 32, left gonocoxa, ventral view, Metromenus protervus; 33, left gonocoxa, ventral view, Metromenus palmae. Dorsal ensiform setae of apical gonocoxite. Figs 23–33, shown with dashed outline. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 142 James K. Liebherr and Elwood C. Zimmerman Figs 34–39. Female bursa copulatrix (bc), spermatheca (sp) with spermathecal gland (sg) and common oviduct (co), ventral view: 34, Anchotefflus gracilis; 35, Barypristus incendiarius; 36, Barypristus rupicola; 37, Colpodes eremita (spermatheca omitted); 38, Colpodes pacificus; 39, Colpodes buxtoni. Dorsal bursal pouches (Figs 37, 39) and basal bursal pouches (Figs 38, 39) stippled. Figs 40–52. Male aedeagus, right lateral view, with detail of aedeagal apex, euventral view: 40, nsp. ‘mandibularis’; 41, Colpocaccus posticatus; 42, C. lanaiensis; 43, Metromenus palmae; 44, nsp. ‘huhula’; 45, Platynus ambiens; 46, Barypristus incendiarius;. 47, Mysticomenus mysticus; 48, Deropristus puncticeps; 49, nsp. ‘bryophila’; 50, nsp. ‘hihia’; 51, Disenochus aterrimus; 52, D. erythropus. Tip of median lobe extending beyond opening of internal sac stippled. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 143 144 James K. Liebherr and Elwood C. Zimmerman Figs 53–56. Male aedeagus, euventral view: 53, Disenochus brevipes; 54, Baryneus sharpi; 55, Colpodes buchannani; 56, Chalcomenus costatus. 6–7. Apical gonocoxite triangular (0,0) (Figs 31–33); gonocoxite subparallel (1,0) (Figs 29, 30); gonocoxite narrow, elongate (1,1) (Fig. 28). 8. Gonocoxal medial margin straight to evenly curved (0) (Fig. 31); medial margin more strongly curved, nearly angulate (1) (Fig. 33). 9–11. Apical gonocoxite triangular, subparallel or elongate, lateral ensiform gonocoxal setae not restricted to base of gonocoxite (0,0,0) (Figs 24–26, 28–33); gonocoxite lateral edge concave, shape scimitarlike (1,0,0) (Fig. 23); concave lateral edge with expanded cutting margin (1,1,0) (Fig. 27); expanded cutting margin restricting lateral setae to basal 1/ 3 of gonocoxal length (1,1,1) (Fig. 27). 12. Basal gonocoxite with ù three setae at medioapical angle (0) (Figs 25, 31–33); medioapical edge with , three setae (1) (Figs 24, 29). 13. Basal gonocoxite with well-developed lateroapical setae (0) (Figs 27, 30–33); lateroapical setae reduced in size, peglike (1) (Figs 28, 29). 14. Basal gonocoxite with two to four lateroapical setae (0) (Figs 23–33); 0–1 lateroapical setae (1). 15–17. Apical gonocoxite nematiform setae ø 0.303 length segment (0,0,0) (Figs 28, 31–33); setae ù 0.34, ø 0.523 length segment (1,0,0) (Figs 23, 24, 26, 30); setae ù 0.56, ø 0.723 length segment (1,1,0) (Figs 27, 29); setae ù 0.763 length segment (1,1,1) (Fig. 25). The longer of the two nematiform setae was measured from its articulation in the apical furrow to its apex. If the seta was curved, the ocular micrometer was placed against the median point of the curve, and rotated to assess the straight-line length of the seta. The apical gonocoxite was measured as the straight-line distance from the mediobasal articulatory condyle to the tip. 18–19. Bursa copulatrix bilaterally symmetrical (0,0) (Figs 7, 20–22, 37–39); bursa with large left lobe (1,0) (Figs 34, 35); bursa with large right lobe (?,1) (Fig. 36). Barypristus rupicola and Prodisenochus terebratus both exhibit a bursal expansion on the right side (character 19). In order to test whether this condition is synapomorphous, derived from the primitive condition of a symmetrical bursa, or derived via a left-right symmetry switch, character 18 was coded as ambiguous for these taxa. This decoupled characters 18 and 19 for these two taxa, allowing the weight of other characters to define the sister taxa to these species. 20–21. Bursa copulatrix with median lumenal band of spikelike microspicules (0,0) (Fig. 11); bursa with band of mixed spikelike and cristate microspicules (1,0); bursa with band of cristate microspicules (1,1) (Figs 12–14, 38). 22. Bursal microspicules dense (0); microspicules sparse (1). 23. Dorsal bursal pouch absent (0); dorsal bursal pouch present (1) (Figs 15–19, 37, 39). 24. Dorsal bursal pouch broad, rounded, unsclerotized (0) (Fig. 16); pouch narrow, or keyhole shaped (1) (Fig. 18). 25–27. Dorsal bursal pouch broad, rounded unsclerotized (0,0,0) (Fig. 20); pouch with lateral extensions, ‘winged’ (1,0,0) (Fig. 21); pouch with lateral extensions as well as basal extensions that extend to near gonopore (1,1,0); pouch © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini 145 Figs 57–75. External cuticular structures used in cladistic analysis: 57, Colpodes buchannani, dorsal view; 58, labrum, dorsal view, C. buchannani; 59, labrum, dorsal view, Chalcomenus costatus; 60, labrum, dorsal view, Disenochus anomalus; 61, labrum, dorsal view, Atrachycnemis sharpi; 62, pedicel, apical view, C. buchannani; 63, pedicel, dorsal view, Colpocaccus posticatus; 64, pedicel, apical view, C. posticatus; 65, pedicel, apical view, D. munroi; 66, pedicel, apical view, D. curtipes; 67, vertex, Chalcomenus molokaiensis; 68, vertex, Atelothrus transiens; 69, vertex, Atelothrus howarthi; 70, pronotum, C. posticatus; 71, pronotum, Mesothriscus vagans; 72, pronotum, Metromenus calathoides; 73, left elytron, Mesothriscus alternans; 74, left elytron, Deropristus puncticeps; 75, left elytron, Atrachycnemis sharpi. with lateral extensions very long, extending across dorsum of bursa (1,0,1) (Fig. 22). 28. Dorsal bursal pouch unsclerotized (0); pouch sclerotized, with microsculpture (1) (Fig. 15). 29. Dorsal bursal pouch broadly rounded (0); pouch truncate (1) (Fig. 17). 30. Dorsal bursal pouch unsclerotized (0); pouch sclerotized in liplike or collarlike shape (1) (Fig. 19). 31. Basal bursal pouch absent (0); basal pouch present (1) (Figs 38, 39). 32. Basal pouch broadly rounded, unsclerotized (0); pouch heavily sclerotized, keyhole or pocket shaped (1) (Fig. 38). For characters 24–32, if a taxon lacks the dorsal bursal pouch or the basal bursal pouch, the taxon is coded 0 for any modification of that pouch. 33–34. Bursa copulatrix subequal to 1.53 length common oviduct (0,0); bursa ù 1.6, ø 2.53 length common oviduct © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 (1,0) (Figs 34–37, 39); bursa ù 2.63, to 43 length common oviduct (1,1) (Fig. 38). After 24 h cold KOH treatment, the pedicel of the ovary dissolves beyond the sclerotized cuticle of the common oviduct-lateral oviducts junction. Common oviduct length was measured from the median junction of the two lateral oviducts to the base of the spermathecal duct. Bursa copulatrix length was measured after the bursa was extended, and was the distance from a point medial to the basal processes of the basal gonocoxites to the spermathecal ductcommon oviduct junction. If the bursa was curved after extension, a line following the curve was used to determine bursal length. 35. Apical bursal pouch absent (0); apical pouch present (1) (Fig. 7). 36. Lateral bursal glands absent (0); bursal glands present (1) (Fig. 7). 146 James K. Liebherr and Elwood C. Zimmerman © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini 37–38. Spermatheca without basal lobe (0,0) (Figs 7, 8); with small basal lobe (1,0) (Fig. 9); with large basal lobe (1,1) (Fig. 10). 39. Spermathecal reservoir fusiform (0) (Fig. 7); reservoir with adductal bulge basally (1) (Fig. 8). 40. Spermatheca without basal digitiform process (0); basal digitiform process present (1). 147 41. Spermathecal reservoir with 12–30 constrictions (0) (Figs 8–10, 34); larger constrictions lacking, . 40 to µ70 fine sclerotized rings on reservoir (1) (Figs 7, 35, 36). 42. Spermathecal reservoir apical (0) (Figs 7–10, 34–39); reservoir basal, an apical filament present (1). 43. Vaginal setae not pediculate (0); vaginal setae pediculate (1). Fig. 86. Strict consensus cladogram for outgroups; Hawaiian ingroup at node 335. Taxon numbers key to numbers in data matrix (Appendix 1); taxon and node numbers key to list of unambiguous character state changes (Appendix 2); character support indicated by numbers under nodes. Geographic distributions are bracketed. Area abbreviations: Asia (As), Australia (Au), Fiji (Fi), Jamaica (Jam), Japan (Jap), Java (Jav), Mexico (Mx), New Guinea (NG), New Zealand (NZ), Philippines (Ph), South America (SA), Samoa (Sam), Solomon Islands (Sol), Tahiti (Tah), Vanuatu (Van), geographical locality unknown (?). Species followed by MNHP coden are undescribed species in Paris Museum from Vanuatu. Figs 76–85. Scanning electron micrographs of metatarsomeres: 76, right metatarsomeres 2–5, dorsal view, Mesothriscus alternans; 77, right metatarsomeres 2–5, dorsal view, Colpocaccus posticatus; 78, right metatarsomeres 2–5, dorsal view, Barypristus rupicola; 79, metatarsomeres 3–5, dorsal view, Metromenus caliginosus; 80, right metatarsomeres 2–5, dorsal view, nsp. ‘hihia’; 81, left metatarsomeres 2–5, ventral view, M. alternans; 82, left metatarsomeres 2–5, ventral view, M. caliginosus; 83, left metatarsomeres 2–5, ventral view, B. rupicola; 84, left metatarsomeres 2–5, ventral view, Atelothrus dyscoleus; 85, left metatarsomeres 3–4, ventral view, Colpodiscus lucipetens. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 148 James K. Liebherr and Elwood C. Zimmerman 44. Bursa copulatrix membranous (0); bursal walls heavily sclerotized, ‘leathery’ (1). the aedeagal internal sac. Depth of the median lobe tip is measured at this junction of sac and sclerotized median lobe tip. 57. Tip of aedeagal median lobe not hooked (0) (Figs 40– 45, 48–52); tip hooked ventrally (1) (Figs 46, 47). Male genitalic characters The aedeagus rests in a position rotated 90° clockwise from the exerted position; therefore, the right paramere is ventral in resting position. As a symmetrical aedeagus is plesiomorphic among Carabidae, the orientation of the exerted aedeagus is used for describing various facies of the structure. The terms eudorsal and euventral are used to describe the dorsum and venter in the symmetrical exerted position. Aedeagal median lobe, euventral view 58–62. Apical half of aedeagal median lobe evenly narrowed to tip (0,0,0,0,0) (Figs 40–44, 50–52); apex broad, shovel-like (1,0,0,0,0) (Fig. 48); apex slightly pinched laterally before tip (0,1,0,0,0) (Figs 45, 47, 49); apex attenuate, more strongly pinched laterally before tip (0,1,1,0,0) (Fig. 53); apex strongly attenuate, tip nipplelike (0,1,1,1,0) (Figs 46, 54); apex strongly attenuate and elongate, tip chisellike (0,1,1,1,1) (Fig. 55). Aedeagal median lobe, lateral view 45. Aedeagal median lobe gracile (0) (Figs 40–44, 47, 49, 51); shaft stout (1) (Figs 45, 46, 48, 50, 52). Whether the median lobe was gracile or stout was determined by the ratio a/b, of (a) the greatest depth of the lobe (measured dorsoventrally) to (b) the straight-line distance from the tip of the lobe to the junction, on the eudorsal surface, of the basal bulb and aedeagal shaft. A ratio of 0.19 or less was scored gracile, and a ratio of 0.20 or more was scored robust. 46–48. Aedeagal median lobe evenly curved on euventral surface (0,0,0) (Fig. 40); straight (1,0,0) (Figs 41, 43–47, 51, 52); recurved (1,1,0) (Figs 42, 50); strongly recurved (1,1,1) (Fig. 48). Character 48 is currently interpreted as an autapomorphy of Deropristus puncticeps. However, males of the closely related D. blaptoides and nsp. ‘kipahulu’ have not been examined because of the lack of males for the former, and the teneral status of the single known specimen of the latter. Including this character allows prediction of states for those taxa. 49. Aedeagal median lobe without euventral excavation (0) (Fig. 40); euventral excavation present (1) (Figs 41–52). The euventral excavation is an invaginated groove on the ventral surface of the median lobe shaft just basad the apex. 50–52. Apex of aedeagal median lobe evenly curved (0,0,0) (Figs 40–42, 44–47, 52); apex angled ventrally (1,0,0) (Fig. 49); apex strongly angled ventrally (1,1,0) (Figs 48, 50, 51); apex not angled, straight ventrally (0,0,1) (Fig. 43). Aedeagal median lobe tip, lateral view 53. Tip of aedeagal median lobe tightly rounded (0) (Figs 40– 46, 50–52); tip acuminate (1) (Figs 47–49). 54. Tip of aedeagal median lobe not downturned (0) (Figs 40–45, 47, 50, 52); tip downturned (1) (Figs 46, 48, 49, 51). 55. Tip of aedeagal median lobe tightly rounded to acuminate (0) (Figs 40–45, 48–52); tip broader, forming a ‘bottlenose’ (1). 56. Tip of aedeagal median lobe , 2.03 long as deep (0) (Figs 40–47, 49–52); tip ù 2.03 long as deep (1) (Fig. 48). The length of the median lobe tip is measured as the distance from the aedeagal tip to its eudorsal junction with the base of Aedeagal median lobe shaft 63. Shaft of aedeagal median lobe evenly narrowed from base to apex, euventral view (0); shaft constricted medially, slightly expanded before attenuation of apex (1) (Fig. 56). 64. Shaft of aedeagal median lobe of moderate breadth, euventral view (0) (e.g. Fig. 43); shaft narrow, needlelike in euventral view (1) (Fig. 44). 65. Shaft of aedeagal median lobe of normal depth in lateral view (0); shaft slender (1) (Figs 41, 42). 66. Eudorsal surface of aedeagal median lobe uniformly curved (0); surface dorsally expanded at midpoint of length (1) (Fig. 43). 67. Aedeagal sac surface spineless (0); sac with spines (1) (Fig. 55). Spination of the internal sac is restricted to outgroup taxa in this analysis. We have not attempted to homologize specific fields of aedeagal spination, but use the simpler presence/ absence coding because of the relatively sparse taxon sampling among the outgroup taxa, especially of the extremely speciose genus ‘Colpodes’ (5 Platynus Bonelli) (Liebherr, 1992, 1998). External characters 68–70. Ocular ratio (OR) 1.38 ø OR ø 1.61 (0,0,0) (Fig. 67); OR ù 1.63 (1,0,0) (Fig. 57); 1.23 ø OR ø 1.36 (0,1,0) (Fig. 68); 1.00 ø OR ø 1.20 (0,1,1) (Fig. 69). The ocular ratio (OR) is the ratio of the maximum width across the compound eyes divided by the minimum width between the compound eyes. The OR was first determined for all ingroup and outgroup taxa, using as many as three individuals if available. The four character states were delimited by gaps in the distribution of observed ratios. 71. Eye diameter large relative to depth (0) (Figs 57, 67– 69); eye diameter small but eyes strongly protruding, bugeyed (1). 72. Six labral apical setae (0) (Figs 57–61); variably 4–5, or 4 setae (1). 73–75. Labral apex straight medially (0,0,0) (Fig. 58); apex broadly, shallowly emarginate (1,0,0) (Fig. 59); apex broadly, © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini 149 Fig. 87. Strict consensus cladogram for Hawaiian ingroup taxa, including nsp. ‘mandibularis’, Colpocaccus species, and species of Sharp’s (1903) Division 1. Division 1 is sister group to a revised Division 2 (node 332). Conventions as in Fig. 86. Area abbreviations: Kauai (K), Oahu (O), Molokai (Mk), Lanai (L), West Maui (WM), East Maui or Haleakala (EM), Hawaii or Big Island (H). © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 150 James K. Liebherr and Elwood C. Zimmerman Fig. 88. Strict consensus cladogram for Hawaiian ingroup taxa of Division 2. Conventions as in Figs 86, 87. Sketch map of islands shows relative sizes of Kauai (K), Oahu (O), Molokai (Mk), Lanai (L), West Maui (WM), East Maui (EM) and Hawaii (H). Relative island positions modified to fit figure. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini moderately emarginate (1,1,0) (Fig. 60); apex with rounded lobes laterally and narrow median notch (1,1,1) (Fig. 61). 76–78. Frons microsculpture isodiametric, sculpticells shiny (0,0,0); granulate isodiametric microsculpture, surface not shiny, leathery (1,0,0); shiny, weak isodiametric sculpticells (0,1,0); sculpticells reduced, surface shiny (0,1,1). 79. Neck with strong dorsal impression (0); slight dorsal impression (1). 80. Third antennomere glabrous except for apical ring of setae (0) (Fig. 57); sparsely setose over surface (1). 81–83. Pedicel with one outer apical seta (0,0,0) (Fig. 62); outer seta plus one or two apical setae (1,0,0) (Figs 63, 64); outer seta plus three to five apical setae (1,1,0); outer seta 1 ù six apical setae (1,1,1) (Figs 65, 66). Fig. 89. Resolved cladogram for species of Division 1 used for biogeographic analysis. Total path length from common ancestor shared with Division 2 is 602 steps under slow, i.e. deltran optimization. Area relationships for more basal taxa – nsp. ‘mandibularis’ and Colpocaccus species, can be derived from Fig. 87. Area abbreviations: Kauai (K), Oahu (O), Molokai (Mk), Lanai (L), West Maui (WM), East Maui (EM) and Hawaii (H). Fig. 90. Resolved cladogram for species of revised Division 2 used in biogeographic analysis. Total path length from common ancestor shared with Division 1 is 448 steps under slow optimization. Area abbreviations as in Fig. 89. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 151 152 James K. Liebherr and Elwood C. Zimmerman Figs 91–94. Four area cladograms (Trees A–D) that minimize duplications of the COMPONENT tree mapping algorithm. 84. Pedicel apical setae short (0) (Figs 64, 66); apical setae almost as long as outer seta (1) (Fig. 65). 85–86. Mandibles moderate, acuminate (0,0) (Fig. 57); elongate (1,0); stout, foreshortened (0,1) (Fig. 3). 87. Two supraorbital setae present (0); one supraorbital seta, anterior seta absent (1). 88. Two submentum setae each side, or four to seven each side (nsp. ‘mandibularis’ autapomorphy) (0); one seta each side, or unilaterally one (1). The autapomorphic condition of nsp. ‘mandibularis’ was considered synonymous with the plesiomorphic state of two setae on each side for this analysis. Should another outgroup or Hawaiian ingroup taxon be found with four or more setae on each side of the submentum, the states of this character would need to be reassessed. 89–90. Mentum median projection triangular to rounded apically (0,0); apex truncate to slightly bifid (1,0); apex bifid or broadly bifid (1,1). 91–96. Pronotal marginal gutter broad, edge upturned (0,0,0,0,0,0) (Figs 6, 57, 70, 72); margin very broad, edge upturned (1,0,0,0,0,0); marginal gutter moderate, edge upturned (0,1,0,0,0,0) (Figs 1, 5, 71); marginal gutter moderate, edge beaded (0,1,1,0,0,0); marginal gutter narrow, edge upturned (0,1,0,1,0,0,0); marginal gutter narrow, edge beaded (0,1,1,1,0,0) (Fig. 4); marginal gutter obsolete, marginal bead present (0,1,1,1,1,0) (Figs 2, 3); marginal gutter and bead absent, external ridge scarcely visible to absent (0,1,1,1,1,1). The derived states for characters 93–94 imply a reticulate character state transformation series, with presence or absence of a marginal bead, and a moderate or narrow marginal gutter co-occurring in all possible combinations. As all combinations of states are observed, the cladistic analysis serves to determine specific hypotheses of character evolution for taxa possessing derived and secondarily reversed states of these characters. 97. Anterior pronotal seta present (0) (Figs 57, 71); absent (1) (Figs 70, 72); polymorphic (?). 98. Posterior pronotal seta present (0) (Figs 57, 70); absent (1) (Figs 71, 72); polymorphic (?). 99–100. Pronotal basal bead complete (0,0) (Figs 57, 70– 72); bead effaced medially (1,0) (Fig. 2); bead effaced medially and laterally (1,1) (Figs 3, 4). Taxa polymorphic for pronotal basal bead development are coded ambiguously for either character 99 or 100. 101–102. Pronotal median base smooth or longitudinally wrinkled (0,0) (Figs 1, 2, 5, 6, 57, 70–72); punctate (1,0); large pits present, surrounded by smooth cuticle (1,1) (Figs 3, 4). 103–104. Pronotal basolateral margin straight (0,0); margin expanded posteriorly, basal margin concave to trisinuate (1,0) (Figs 6, 57, 72); margin concave, basal margin expanded medially, sinuate laterally (0,1). 105–110. Pronotal laterobasal depressions broad, smooth, evenly concave (0,0,0,0,0,0) (Figs 1, 57); depressions broad, smooth, with median tubercle (1,0,0,0,0,0) (Figs 5, 6); depressions linear, smooth, with broad lateral convexity (0,1,0,0,0,0); depressions broad, punctate (0,0,1,0,0,0); depressions linear, punctate, with broad lateral convexity (0,1,1,0,0,0); depressions deep, pitlike, punctate (0,0,1,1,0,0); depressions shallow to planar, sparsely punctate (0,0,1,0,1,0) (Fig. 2); depressions obsolete, obscured by large punctures over base (0,0,1,0,0,1) (Figs 3, 4). Characters 106–107 co-occur in all of the four possible combinations. The cladistic analysis forms the means to test the primary homology statements for these two characters, as the distribution of states ensures that homoplasy exists in the character transformation series for this suite of characters. 111–113. Pronotal hind angles obtuse-angulate (0,0,0) (Figs 57, 72); hind angles sharp, right or nearly so (1,0,0) (Figs 3, 6); hind angles obtuse-rounded (0,1,0) (Figs 1, 4, 5, 70, 71); hind angles rounded, nearly obsolete (0,1,1) (Fig. 2). 114–115. Pronotal basolateral margins sinuate before hind angles (0,0) (Figs 3–5, 57); margins straight (1,0) (Fig. 6); margins convex (1,1) (Figs 1, 2). 116–118. Pronotal discal microsculpture transverse mesh (0,0,0); microsculpture strong to granulate transverse mesh (1,0,0); microsculpture reduced transverse mesh (0,1,0); microsculpture obsolete, surface shiny (0,1,1). 119. Pronotal laterobasal depression with transverse mesh microsculpture (0); depression with isodiametric mesh (1). 120. Pronotal disc without large pits (0) (Figs 1, 2, 5, 6, 57, 70–72); disc with large pits over surface (1) (Figs 3, 4). 121. Prosternal process unmargined (0); dorsal triangle of posterior face, mediodorsal to coxae, carinate (1). The dorsal triangle of the prosternal process is defined by the posterior margins of the procoxal cavities and the posterior margin of the prosternum. This area is usually rounded in Carabidae, including most Platynini, but in some Colpodes species (of authors) it is strongly carinate. 122. Prosternal process glabrous (0); apically setose (1). 123. Prosternal process unmargined ventrally (0); ventral margin carinate (1). 124. Prosternal process median ventral surface convex to flat (0); median ventral surface depressed (1). In the derived condition for this character, the median ventral surface of the prosternal process is concave, being broadly invaginated between the posteroventral portions of the procoxae. 125. Prosternal process ventral surface smooth (0); ventral surface punctate (1). 126. Pronotal hind seta at hind angle (0) (Fig. 57); seta 0.02– 0.213 median pronotal length before hind angle (1) (Fig. 70). In all outgroup taxa, the posterior pronotal seta, if present, is located at the hind angles of the pronotum. Some Hawaiian species retain this plesiomorphic condition; Apteromesus © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini maculatus, Deropristus blaptoides, Blackburnia insignis, Colpodiscus lucipetens, Colpocaccus tantalus, and Colpocaccus hawaiiensis. In the other species the posterior setae may be anywhere from 0.02 to 0.213 the length of the pronotum before the hind angles (defined as that posterior point of the pronotal margin exhibiting the strongest curvature). This character was not scored for species lacking hind pronotal setae; however, for species polymorphic for these setae, individuals possessing the setae were used for scoring. 127. Scutellar seta of elytra present (0) (Figs 57, 73–75); seta absent (1). 128–130. Dorsal elytral setae three or three to four (0,0,0) (Figs 57, 73, 74); dorsal elytral setae absent (1,0,0) (Fig. 75); anterior seta absent, two unilaterally or on both sides (0,1,0); four to five dorsal elytral setae (0,0,1). Prodisenochus terebratus varies in presence of the median dorsal seta, with several specimens unilaterally expressing this seta. This species was considered to represent the derived state for character 128. 131. Dorsal elytral setae elongate when present, longer than width of third elytral interval (0); setae reduced in length, peglike, shorter than width of elytral interval (1); dorsal elytral setae absent (?). In the derived state for this character, the dorsal elytral setae are uniformly shorter than the width of the third elytral interval. This condition cannot be attributed to wear, as the setae are uniformly short in all individuals. 132. Dorsal elytral setae not set in foveae (0); setae set in foveae (1); setae absent (0). Foveae were considered depressions that extended at least across the width of the third elytral interval. 133–134. Elytral subapical sinuation present (0,0) (Figs 57, 73); sinuation reduced, margin straight or slightly sinuate at plica (1,0) (Figs 2, 4, 74); sinuation obsolete, margin convex (1,1) (Figs 3, 75). 135–136. Sutural apex rounded, non-denticulate (0,0) (Figs 1–6, 73–75); short tooth present (1,0) (Fig. 57); longer spine present (1,1). 137–138. Subapical tooth absent (0,0) (Figs 1–6, 73–75); subapical prominence present (1,0) (Fig. 57); subapical tooth present (1,1). 139. Elytral striae smooth, continuous (0) (Figs 1, 5, 6, 73); striae punctulate to punctate at least partially (1) (Figs 2–4, 74, 75). 140. Elytral punctures absent to pitlike, striae continuous, or if reduced, pits connected by depressed areas (0) (Fig. 4); punctures pitlike, striae discontinuous, usually at least in apical half, sometimes more so (1) (Fig. 3). 141–142. Elytral punctures absent to fine, punctulate (0,0); punctures moderate, expanding striae laterally (1,0) (Fig. 3); punctures large, pitlike, greatly expanding striae laterally (1,1) (Fig. 4). 143. Elytral striae evenly punctured throughout length (0) (Fig. 2); punctures strongest in basal 1/2 to 2/3 of length, reduced apically with striae not reduced apically (1) (Fig. 4). Scoring the derived state of this character required the absence of punctures in striae near the apex of the elytra, and © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 153 fully developed striae. If the striae and associated punctures were both absent apically, character 145 was scored. 144. Elytral striae equally developed laterally (0) (Figs 3, 4); striae laterally evanescent (1) (Fig. 2). 145. Elytral striae evident apically (0) (Figs 3, 4); striae apically evanescent (1) (Fig. 2). 146–150. Elytral intervals without costae (0,0,0,0,0) (Figs 2, 4, 6, 74, 75); weakly rounded costae on intervals 1, 5 and 7, or intervals 1, 3, 5 and 7 (1,0,0,0,0); larger angulate costae on intervals 1, 3, 5 and 7, and smaller costae on intervals 2, 4, 6 and 8 (1,1,0,0,0) (Figs 5, 73); smaller rounded to angulate costae on intervals 1, 3, 5, 6 and 7 (1,0,1,0,0) (Fig. 1); angulate costae on intervals 1, 5, 7 and rounded costa on 6 (1,0,0,1,0); angulate costae on intervals 1, 3, 5, 6, 7 and 8 (1,0,0,1,1). Costate elytral intervals are observed only in Kauai species of this radiation, with the single exception of Anchotefflus gracilis of Oahu. Within any taxon, the strongest costae are present on odd-numbered intervals; 1, 5 and 7, or 1, 3, 5 and 7. If these intervals are only weakly costate, no other costae are present, but if costae on the odd intervals are large, rounded weaker costae are present on the even intervals. This may involve only interval 6, or intervals 2, 4, 6 and 8. In the above transformation series, rounded costae are assumed to be more plesiomorphic than larger angulate costae. Characters 147–149 are considered unordered relative to character 146. Character 148, coding 1,0,1,0,0, is an autapomorphy of Apteromesus maculatus. It was included to investigate whether the restricted development of rounded costae on interval 6 is evolutionarily associated with presence of rounded costae on all even intervals (character 147). Character 150 is a derivation of character 149. 151. Elytral intervals slightly convex to costate (0) (Figs 1– 6); intervals nearly flat (1) (Fig. 57). 152. Elytral intervals slightly convex to costate (0); intervals broadly convex (1). 153–155. Humeri broad (0,0,0) (Figs 4, 57, 75); humeri broad and quadrate, ‘abacoid’, lateral margin expanded (1,0,0) (Fig. 6); humeri moderate, slightly narrowed (0,1,0) (Figs 1– 3, 5, 73, 74); humeri reduced, very narrow (0,1,1). The humeri are highly modified in the various brachypterous Hawaiian taxa. The usual syndrome for wingless taxa is the reduction of the humeri, resulting in a pedunculate body shape (Fig. 2). In some Hawaiian taxa, however, the opposite trend is observed; the humeri become broadly flanged. This broadening is often associated with a quadrate pronotum, so that the broad humeri continue a posteriad expansion of the body due to a quadrate or trapezoidal pronotum (Fig. 6). As this body shape is reminiscent of the pterostichine genus Abax Bonelli, we describe the shape as abacoid. 156–157. Humeral angle rounded (0,0) (Figs 2–4, 57); humeri tightly rounded (1,0) (Fig. 1); humeri angulate (1,1) (Figs 5, 6, 73). 158. Basal humeral groove present, straight to evenly curved (0) (Figs 1, 5, 6, 57, 73); humeral groove obsolete, difficult to trace laterad scutellum (1) (Figs 3, 74, 75). 159–161. Elytral disc moderately flat, depressed laterally (0,0,0) (Figs 1, 5, 6, 57); elytra flattened medially (1,0,0); elytra convex, shaped as a ship’s hull, strongly depressed 154 James K. Liebherr and Elwood C. Zimmerman laterally and apically (0,1,0) (Figs 2–4); elytra domelike, disc vaulted medially, elevated above level of scutellum (0,0,1). Three major trends in elytral shape evolution are observable: (1) the elytral disc is flat in species associated with leaf axils, much as seen in the bromeliad inhabiting Jamaican Platynus (Liebherr, 1988), (2) the elytra of brachypterous Division 1 taxa may be fused at the suture, and depressed laterally and apically (Figs 2–4), and (3) the elytra of some brachypterous Division 2 species may become elevated caudad the scutellum, giving the elytra a vaulted appearance. 162–163. Sixteen to thirty-two lateral elytral setae (0,0) (Fig. 57); twelve to sixteen setae (1,0); eight to twelve setae (1,1) (Fig. 75) (no taxon monomorphically displays twelve setae; taxa that span state limits coded ? for that character). The lateral elytral setae comprise all setae occurring in or near the eighth elytral stria, before that stria reaches the lateral margin of the elytra. The subapical setae associated with interval 7 and the apical seta associated with interval 1 (Fig. 57) are thus excluded from this setal category. 164–165. Elytra ferruginous to piceous (0,0) (Figs 2–6); elytra an irregular mosaic of flavous and brunneous flecks (1,0) (Fig. 1); elytra flavous overall (0,1). 166–168. Elytra unicolourous or a mosaic of flecks (0,0,0) (Figs 1–4, 6); apex flavous, disc and lateral margins ferruginous to piceous (1,0,0); disc darker, lateral margins and apex flavous (0,1,0) (Fig. 5); disc flavous, sides and apex darker (0,0,1). The elytra of Apteromesus maculatus, Mysticomenus mysticus and M. tibialis all exhibit a highly unusual irregular mosaic of melanized and pale cuticle. The pattern of melanization does not match from individual to individual, nor from side to side within individuals. The only other taxa in which we have observed this condition are the ‘poxed’ species of Maculagonum Darlington, resident on New Guinea, e.g. Maculagonum seripox, an outgroup for this analysis. The condition of flavous elytral disc with darker sides and apex (character 168) is an autapomorphy of Metromenus mutabilis, and polymorphic within M. caliginosus. 169–172. Elytral microsculpture transverse mesh 2.0–3.03 wide as long (0,0,0,0); microsculpture fine transverse lines without mesh or mesh . 3.03 wide as long (1,0,0,0); microsculpture isodiametric mesh in transverse rows (0,1,0,0); microsculpture isodiametric mesh (0,1,1,0); microsculpture granulate isodiametric mesh in transverse rows (0,1,0,1); microsculpture granulate isodiametric mesh (0,1,1,1); microsculpture absent (–,–,–,0). If elytral microsculpture is so reduced as to preclude determination of its shape, we felt unable to determine whether such loss was derived via reduction of isodiametric or transverse sculpticells. However, we believe such a condition cannot be derived via reduction from a heavy granulate microsculpture, as such a condition would necessarily pass through a less granulate state on the way to absence. Thus character 172 was scored 0 for taxa with absent microsculpture. The derived state for such absence was encoded in character 174 below. We also did not want to constrain how granulate isodiametric microsculpture might come about. Therefore states of characters 171–172 permit granulate isodiametric sculpticells to arise via roughening of isodiametric sculpticells not arranged in rows (states 1,0 transform to 1,1), or via shortening of roughened isodiametric sculpticells arranged in rows (states 0,1 transform to 1,1). 173–174. Elytral microsculpture present (0,0); microsculpture reduced but traceable (1,0); microsculpture obsolete (1,1). 175. Basal abdominal ventrites smooth or wrinkled laterally (0); basal ventrites distinctly punctate, from one to many pitlike punctures present (1). 176. Visible abdominal ventrites 3 to 5 not convex, sutures not strongly impressed (0); abdominal ventrites 3 to 5 strongly convex posteriorly, sutures deeply impressed (1). 177–178. Apical female abdominal setae two each side (0,0); female abdominal setae unilaterally or bilaterally three, total of five to six setae, or bilaterally five for a total of ten setae (autapomorphy of nsp. ‘mandibularis’) (1,0); bilaterally four for a total of eight setae (0,1). As in the setation of the mentum, nsp. ‘mandibularis’ is autapomorphic for the number of female apical abdominal setae. We resisted defining a transformation series defined by a linear increase in the number of female setae, as doubling of setal number is as likely an explanation of setal increase as incremental increase. If another taxon is found that possesses five female setae each side, this character will need to be revisited. 179. Profemur with 0 or 1 anteroventral setae (0); femur with two anteroventral setae (1). 180. Mesofemur with two or unilaterally three anteroventral setae (0); femur with from three to seven setae each side (1). 181. Metacoxa with three setae (0); inner seta absent, outer two present (1); inner seta polymorphic for presence (?). 182. Metafemur with one or more dorsoapical setae (0); metafemur dorsoapically glabrous (1) (Fig. 57). 183–184. Metatarsi narrow, gracile (0,0) (Figs 76, 80); tarsi broadened apically (1,0) (Figs 77, 78); broadened overall relative to length (1,1) (Fig. 79). 185. Metatarsi moderately long, basal tarsomere . 23 length of inner tibial spur (0) (Fig. 77); tarsi stout, shortened and of moderate width, basal segment , 2.03 length tibial spur (1) (Fig. 79). 186–187. Ventral vestiture sparse on metatarsomeres 2–4, consisting of three sparse lateral rows of setae each side bordering central space (0,0) (Figs 81, 82); vestiture thicker, densely packed lateral rows of setae bordering central space (1,0) (Fig. 83); ventral surface more densely setose, glabrous central space narrower or obsolete, especially on tarsomere 2 (1,1) (Figs 84, 85). An increase in ventral tarsal vestiture is observed in species associated with plant surfaces. For terrestrial and riparian species, the tarsal venters are usually clothed with three parallel rows of setae. In taxa where these lateral rows are thicker, the space between the rows may become narrower, especially on the basal tarsomeres 1 and 2. For characters 183–187, metatarsi were used where possible for scoring character states, although mesotarsal configurations paralleled those of the metaleg. Male protarsi were studied with light microscopy, but no meaningful variation was observed. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini 188–191. Metatarsomeres 1–3 with moderate inner and outer tarsal grooves (5 dorsal sulci) (0,0,0,0) (Figs 76, 77); tarsi with fine inner and outer sulci (1,0,0,0); tarsi with fine outer and obsolete inner sulci (1,1,0,0); sulci absent, tarsi dorsally convex (1,1,1,0) (Figs 78, 80); deep inner and outer sulci, tarsi medially carinate (0,0,0,1). Sharp (1903) recognized a natural group he called Division 1 of Hawaiian platynines based on their possession of broadly convex tarsomeres. This condition is scored derived for characters 188–190. A derived condition in another direction involves the development of a median tarsal carina based on the presence of extremely deep and broad dorsal sulci. 192–194. Outer lobe of metatarsomere 4 ø length tarsal base, measured from median insertion of metatarsomere 5 to dorsoapical margin of metatarsomere 3 (0,0,0) (Figs 76–78, 80); outer lobe 1.1–1.753 length base (1,0,0) (Fig. 79); outer lobe 2.0–2.753 length base (1,1,0); outer lobe . 3.03 length base (1,1,1) (Fig. 85). Primitively the apex of the fourth tarsomere, scored in this analysis using the metatarsi only, is straight or only slightly emarginate. In the Hawaiian Platynini, this tarsomere may exhibit subequal inner and outer apical lobes, although the outer lobe may still be slightly longer than the inner (Figs 77, 79, 84). This pattern of elongation differs from that observed in the Colpodes of authors, wherein the outer lobe may be much more elongate than the inner lobe; that condition is scored using character 195. In Character evolution (below) we justify this decision based on the results of the cladistic analysis. 195. Outer lobe of fourth metatarsomere subequal to 1.53 length inner lobe (0); outer lobe . 1.53 length inner lobe (1). In the Hawaiian platynines, the outer lobe of the fourth metatarsomere varies from being subequal in length to the inner lobe, to being 1.253 the length of the inner lobe. This character was divided into states based on the outer lobe length being 1.53 the inner lobe, as this ratio distinguished those outgroup taxa with very long outer lobes. In the most extreme cases, the outer lobes are twice as long as the inner, e.g. Colpodes xanthocnemis. 196–198. Fifth metatarsomere (5MT) apparently glabrous ventrally (0,0,0) (Fig. 81); 5MT with four to six setae, shorter than depth of segment measured at insertion point (1,0,0) (Figs 82, 84); 5MT with four to eighteen setae subequal to depth of segment at insertion point (1,1,0); 5MT with eight to eighteen setae 1.53 depth of segment at insertion point (1,1,1) (Fig. 83). The setosity of the fifth tarsomere is often used as a key character for diagnosis of platynine taxa (e.g. Lindroth, 1966; Liebherr & Will, 1996). Apparently glabrous tarsomeres still possess short peglike sensilla that are homologous with longer visible setae (Fig. 81). Rudimentary pectens on the claws, arguable homologous with those observed in platynine genera Calathus Bonelli and Onypterygia Dejean are also observable using SEM (Figs 81, 84). These were not scored in the analysis because we did not conduct SEM analysis using sufficient specimens of all taxa to allow assessment of variation in these structures. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 155 199. Fourth metatarsal (MT4) apical setae located apically (0) (Figs 76, 77); both MT4 apical setae located subapically (1) (Figs 78–80). Gracile tarsomeres exhibit the primitive condition for this character (e.g. Fig. 76). On apically broadened tarsomeres possessing elongate tarsal lobes, the apical setae are often situated some distance basad the tip of the lobe (e.g. Fig. 79). However, this is not always the case, as the broadened tarsomeres of Colpocaccus posticatus (Fig. 77) retain the plesiomorphic placement of the apical seta, thereby justifying the independence of this character. 200–204. Flight wings large, macropterous and functional, wing ratio (WR) 2.9–5.0 (0,0,0,0,0); flight wings shorter, functional, WR 1.4–2.3 (1,0,0,0,0); flight wings brachypterous, not functional, apex strongly shortened, WR 0.4–1.1 (1,1,0,0,0); flight wings reduced both in length and in width, the wing rudiment a long narrow strap or whip, WR 0.4–1.1 (1,1,1,0,0); flight wings vestigial, extending slightly beyond metanotum (1,1,0,1,0); flight wings vestigial, not extending beyond metanotum (1,1,0,1,1). The wing ratio (WR) is defined as the metathoracic alar length (measured from costal vein base to wing apex) 3 maximum alar width, divided by elytral length (defined as distance from scutellar apex to elytral sutural apex) 3 maximum elytral width. Elytral measurements were made on the left elytron. Taxa exhibiting the plesiomorphic states for characters 200–204 undoubtedly fly at times, as specimens have been recorded captured at light (unpublished data). Wing reduction has occurred along two axes among the taxa in this analysis: (1) reduction in wing length with little or at most proportional decrease in wing width and (2) strong reduction in wing width, with retention of a long whiplike wing rudiment composed of the leading edge of the wing (observed in nsp. ‘apterum’ of Fiji, nsp. ‘agonoides’ of unknown provenance and Bryanites samoaensis of Samoa. These two conditions are considered homologous for characters 200–201, but the whiplike condition is scored derived for character 202, whereas shortening of a still broad wing remnant is scored derived for characters 203– 204. Two species from New Guinea, Nebriagonum cephalum and N. transiens, exhibit vestigial wings that extend slightly beyond the metanotum. Because these species also share other derived character states with the geographically proximate Fijian nsp. ‘apterum’ and Samoan B. samoaensis, they were coded 1,1,?,?,0 for wing configuration, allowing the cladistic analysis to determine whether their vestigial wings have evolved through shortening of a broad alar expanse as in the Hawaiian taxa, or via shortening of a whiplike extension, as in the Fijian and Samoan brachypterous taxa. 205. Body length 4.6–10.9 mm (0); body length 10.9– 16.0 mm (1). Body length was defined as the sum of (1) the distance from the middle of the anterior edge of the labrum to the cervical collar, (2) the length of the pronotum measured along the midline and (3) the length of the elytra measured from the scutellar apex to the tip of the left elytron at the suture. This measurement protocol permitted all specimens to be compared regardless of pinned position. All specimens of all species were scanned for largest and smallest individuals, with questionably 156 James K. Liebherr and Elwood C. Zimmerman equal-sized individuals measured until a size range was established. Cladistic analysis The 206 unit characters of the 170 taxa were analysed under the assumptions of the parsimony criterion, whereby the shortest network incorporating all character state changes connects all of the taxa. This shortest network was found using the computer program NONA (Goloboff, 1995), run under the Windows 95 operating system, using a Pentium processor with 16 MB of RAM. Missing character information and polymorphic characters (scored as ?) and inapplicable characters (scored as –) are treated identically by NONA, i.e. they are treated as an unknown state for that taxon, with known states determining the character optimization on the cladogram. The data matrix (Appendix 1) was analysed using the hold* (hold as many trees as computer RAM will permit), hold/50 (hold 50 trees for the start of each initial tree search sequence), q1 (shortcut for fast collapsing of trees during branch-swapping), mult*200 (conduct tree searches for 200 random taxon entry sequences), max* (tree bisection-reconnection branch swapping), and report (output status of analysis to the screen) options. The hold* command must be given before reading the data matrix to permit holding more than 1000 trees after reading the matrix. All equally parsimonious trees found were summarized using a strict consensus tree, i.e. components present in the strict consensus are present in all trees. Unambiguously assigned character state changes for each node on the strict consensus cladogram were listed using the apo- option. NONA roots the network at the first taxon in the matrix, thereby producing a cladogram. The matrix was set up with Notagonum submetallicum as the most likely outgroup, given its lowest advancement index (Farris et al., 1970) value of 15, i.e. fifteen characters were judged apomorphic relative to a generalized platynine groundplan (Wagner, 1980). The critical issue for this analysis is not the ultimate outgroup of the analysis, but the outgroup or outgroups to the clade or clades comprising Hawaiian Platynini, as expounded by the rooting methodology of Nixon & Carpenter (1993). In this study, cladistic relationships within any Hawaiian ingroup are based on derived character states relative to its outgroup. But the nature of relationships patristically close to the designated root, N. submetallicum, may be more suspect depending on where this entire set of 170 taxa would join a network of more basal platynines. Character state distributions, alternate rootings, and support for nodes of the cladogram were investigated using CLADOS (Nixon, 1995) under slow or deltran optimization (Swofford, 1991). Character support was estimated by collapsing each node and noting the increase in tree length. This method results in values generally less than patristic distance, but greater than Bremer (1988) support values. We report character support for two reasons: (1) to give an objective assessment of support shared by all trees for important nodes useful for classification and (2) to compare relative levels of character support on various portions of the cladogram. For the first purpose, calculation of character support on the strict consensus is appropriate, as all nodes discussed relative to classification are shared by all trees. For the second purpose, character support values for polytomous nodes may give greater values relative to any one of the most parsimonious, therefore most resolved cladograms. In this case, the strict consensus is less appropriate. However, in all cases where we compare nodal support across the strict consensus, we note how polytomous nodes influence the support values. In no case does such bias exaggerate any noted differences in support. It should be remembered that our character support values overestimate the amount of support at any node, as collapsing a node may result in a tree longer than any on another island of trees (Maddison, 1991). As such, our character support can only be interpreted as the relative amount of nodal support for trees in the same island. Biogeographic analysis The definition of areas of endemism was quite straightforward, as 124 of the 128 native species, or 97%, are distributed on a single island (Figs 87, 88). These distributions assume the islands to be Kauai, Oahu, Molokai, Lanai, West Maui, East Maui or Haleakala, and the island of Hawaii, also called Big Island. Thus, the island of Maui is considered as two areas of endemism, even though West and East Maui are connected by a low isthmus. Species totals for the various island areas are: (1) Kauai (twenty-four species), (2) Oahu (thirty-two species), (3) Molokai (twenty-one species), (4) Lanai (five species), (5) West Maui (seventeen species), (6) East Maui or Haleakala (thirty-five species) and (7) Big Island (six species). The four widespread species include: (1) Colpocaccus lanaiensis found on the largest islands comprising the former superisland Maui Nui, i.e. Molokai, Lanai, West and East Maui, (2) Metromenus sphodriformis, found on Molokai, West and East Maui, (3) Chalcomenus molokaiensis, found on Molokai, West and East Maui, and Big Island, and (4) Colpodiscus lucipetens, distributed on Big Island, West and East Maui, Lanai, Molokai and Oahu. For this latter species, only Big Island appears to support long-term resident populations, as collections on the other islands are extremely sporadic, and always involve lowland sites and often are made in flight at lights. The Oahu record was from a storefront light in Honolulu, and was most certainly an adventive record, perhaps associated with commerce (unpublished data). The tree mapping algorithm described by Page (1990) and implemented in COMPONENT (Page, 1993) was used to determine the best fitting area relationships for the observed distributional data. This method constructs a reconciled cladogram by adding hypothetical taxa, called added leaves by Page (1993), to an observed taxon-area cladogram, called an observed tree by Page (1990). The reconciled tree consists of more leaves than the observed tree, and the difference is quantified as the number of items of error required to change the observed tree into the reconciled tree (Nelson & Platnick, 1981), i.e. the number of items of error is two times the number of leaves added to the © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini observed tree. Leaves can be added by duplication of portions of the observed tree; these duplications are the primary criterion minimized by COMPONENT. Duplications may be thought of as past dispersal events that resulted in sympatry amongst two previously allopatric vicariant sister taxa, with subsequent vicariance of the secondarily sympatric taxa resulting in parallel diversification of two sympatric lineages. Duplications may also reflect divergence along ecological axes not accounted for by strictly geographical circumscription of areas of endemism. COMPONENT also reports a third criterion for determining fit of a pattern of relationships to an observed tree; the number of losses of lineages within areas. The number of losses is related to the number of added leaves, as it is the number of monophyletic sets of added leaves required to achieve fit of an observed cladogram to a user defined hypothesis of area relationships; as such, number of losses is usually less than, although it may be equal to, the number of added leaves. In a host-parasite system, an hypothesized loss would represent loss of a primordial parasite in a primordial host, thereby ensuring the absence of any descendants of the parasite in descendants of the host (Page, 1990) a desirable result in a system where dispersal is discounted and parallel cladogenesis is favoured. In a biogeographic system such as the Hawaiian islands, where losses of lineages could represent either lack of dispersal of a lineage to another island or extinction of a lineage on an island, the desirability of minimizing the losses of higher level lineages is less clear. We use the number of losses in a subsidiary analysis wherein we compare the fit of taxon-area relationships between two major clades of the platynine radiation. In this specific instance, this criterion gives the best estimate of potential extinctions, a parameter we use to estimate potential speciation events, and therefore speciation rate. Whether the taxon-area cladogram deviates from randomness was tested using the random tree generator of COMPONENT (Page, 1993), using a Markovian model for tree generation (Page, 1991). Random area cladograms were fit to the taxonarea cladogram, with duplications minimized. The numbers of duplications, items of error, and losses for these randomly generated area cladograms were then compared to the numbers from the best fitting trees found using the heuristic search. A finding that the heuristic trees are better fitting than 95% or more of the randomly generated trees supported the nonrandom nature of the taxon-area cladogram. Results Cladistic analysis Five of the 200 random taxon entry sequences resulted in discovery of 1651 step trees; The resultant 250 trees were subsequently swapped using the max* tree bisectionreconnection swapping algorithm, with the 16 MB of RAM permitting retention of 14 169 equally parsimonious trees: length 1650, consistency index 0.12, retention index 0.68. In order to assess the likelihood that the strict consensus of the first 14 169 equally most parsimonious trees (Figs 86–88) exhibited resolved relationships that would collapse were all © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 157 trees to be enumerated, strict consensus sets of the first fifty, 100, and 200 trees in memory were constructed. The strict consensus of the first fifty trees was identical to the strict consensus of 14 169 trees, with the following exceptions: (1) Metromenus sphodriformis and Atelothrus depressus were considered to be sister species, (2) Atelothrus constrictus, A. erro, plus A. longulus were considered to be monophyletic and (3) Mesothriscus tricolor and M. concolor were considered to be sister species. The strict consensus of 100 trees differed from the strict consensus of 14 169 trees only by the third component above, and the strict consensus of 200 trees was identical to that reported here (Figs 86–88). That the strict consensus based on 200 trees, or 1.4% of the total number of trees enumerated, is identical to the strict consensus of all enumerated trees supports the interpretation that the many trees beyond 200 reflect the possible combinations of resolved nodes represented by polytomies across the strict consensus cladogram. However, the discovery of more islands of trees (Maddison, 1991) and further deresolution of the strict consensus is certainly possible if more trees were retained. The 1716-step strict consensus cladogram (Figs 86–88) exhibits very little ambiguity among outgroup relationships, and supports monophyly of an all-Hawaiian ingroup (Figs 86, 87, node 335). The taxa of Sharp’s (1903) Division 1 (taxa subtended by node 310, Fig. 87) also retain highly resolved relationships, illustrating the relative lack of ambiguity for character data among those taxa; forty-four of forty-nine components (90%) present in any of the shortest trees are retained in the strict consensus. Relationships among species in a redefined Division 2 (taxa subtended by node 332, Fig. 88) are more ambiguously supported. Of the sixty-eight components supporting resolution among newly defined Division 2 taxa in any of the shortest cladograms, fifty-three (78%) are represented in the strict consensus (Fig. 88). This calculation ignores the node basal to the conspecific terms for Metromenus fraternus. The difference in support is also shown by a comparison of character support across the tree (Figs 87, 88). The support index ranges from 2 to 14 (ignoring the polytomous node 218) for Division 1 (average 5.5), vs. a range of 1–8 (average 2.7) for Division 2. The position of Notagonum submetallicum at the root is based on its low advancement index relative to a platynine groundplan. The two Lorostema species in the analysis are patristic neighbours of N. submetallicum, as well the sister group to the 128 ingroup species of Hawaiian Platynini. We feel it highly unlikely that the root of the cladogram should be placed within the Hawaiian ingroup, given the large number of derived character states defining the monophyly of the Hawaiian radiation (Appendix 2). These include unambiguously placed character state changes for characters 81 (pedicel apically with extra setae, Figs 63–66), 97 (anterior pronotal seta absent, Figs 70, 72), 112 (pronotal hind angles obtusely rounded, Figs 70–72), 126 (pronotal hind seta slightly before hind angle, Fig. 70), 139 (elytral striae smooth, reversal from the punctate state seen in both the sister group Lorostema and the outgroup Notagonum), 177 (females with from three to five setae on each side of apical visible abdominal sternite), and 182 (metafemur dorsoapically glabrous, as in Fig. 57). Denial 158 James K. Liebherr and Elwood C. Zimmerman of Hawaiian platynine monophyly would require an extra step on the cladogram for each of these seven characters (Figs 86, 87). Nonetheless, none of these characters is unique to the Hawaiian ingroup; from one to twenty-seven of the outgroup taxa share the same character state as the base of the ingroup. Moreover, reversed character transformations for all seven characters occur within the ingroup. The possession of one or more extra setae at the apex of the pedicel (character 82) comes closest to being a taxonomically diagnostic character, as only nsp. ‘atra’ lacks such setae. Conversely, the outgroup nsp. ‘apterum’ of Fiji has independently evolved such setae at the apex of the pedicel. For the other characters, four to fortyseven ingroup taxa possess the alternate state from that at the base of the ingroup. These reversals may be rare, as in character 139, wherein all taxa subtended by node 309 except Disenochus curtipes exhibit the derived state, or very common, as in character 112, which requires at least sixteen or more state changes within the ingroup, depending on how the polytomy at node 270 is resolved (the thirteen unambiguously placed changes are listed in Appendix 2). The patristic distance between node 336 subtending the Lorostema species and Notagonum submetallicum is only two steps: transformation from a fusiform female spermatheca to a spermatheca with a series of constrictions (character 41) and reduction from sixteen to thirty-two, to twelve to sixteen, lateral elytral setae (character 162). If we were to place the root of the cladogram such that the Hawaiian ingroup is sister to all other outgroups, we must reverse our interpretation of the above two derivations and consider them to be symplesiomorphic. Such an interpretation does not fit with the presence of a fusiform, non-constricted spermatheca in the subtribe Sphodri, the sister to the subtribe Platyni (Liebherr, 1986), into which all taxa in this study are placed. It also conflicts with the presence of more than sixteen lateral elytral setae in most platynine taxa of both subtribes Platyni and Sphodri. Thus, consideration of Lorostema as the sister group of Hawaiian Platynini is the most parsimonious solution based both on this data set, and more broadly based character-state distributions throughout the tribe. The Hawaiian Platynini are divisible into four monophyletic lineages, all rooted on Kauai: (1) the monotypic sister group to the other 127 species comprising the undescribed species ‘mandibularis’, (2) the four species recognized by Sharp (1903) as the genus Colpocaccus, (3) fifty species allied with species Sharp (1903) recognized as Division 1 and (4) seventy-three species redefining Sharp’s Division 2, excluding the four Colpocaccus species mentioned above. The monotypic ‘mandibularis’ lineage is supported by autapomorphies for characters 4 (broad lateral setae on the apical female gonocoxite, as in Fig. 33), 85 (elongate mandibles), 99 (pronotal bead medially effaced, reversed in its sister group), 186 (tarsomeres with dense lateroventral brush of setae) and 189 (fine outer and obsolete inner dorsal sulci on the basal metatarsomeres). Derived states for characters 4 and 186 are shared with members of the patristically similar genus Colpocaccus, and nsp. ‘mandibularis’ exhibits symplesiomorphic similarity with species of that genus. Nonetheless, a sister group relationship with the other 127 native Hawaiian species supports recognition of this species as a distinct lineage. Monophyly of the genus Colpocaccus is supported by derived states for characters 39 (female spermatheca with an adductal bulge, Fig. 8), and 65 (slender median lobe of male aedeagus, Figs 41, 42). Sharp (1903) included this genus in his Division 2. Based on cladistic criteria, we remove it to its own lineage. Sharp’s (1903) Division 1 is well defined by a number of synapomorphies: derived state changes for characters 6 (female apical gonocoxite subparallel, Figs 25, 26, 28–30), 12 (female basal gonocoxite with , three setae, e.g. Figs 24, 29), 15 (female apical gonocoxite with moderately long apical nematiform setae, Figs 23, 30), 18 (bursa copulatrix asymmetrical, with left lobe, Figs 34, 35), 33 (bursa copulatrix elongate, Figs 34–36), 73 (labrum shallowly emarginate, Fig. 59), 177 (prothoracic femur with none or one anteroventral setae, a reversal to the plesiomorphic state), 189 and 190 (tarsi dorsally convex, Figs 78, 80), and 197 (fifth metatarsomere with four to eighteen moderately elongate ventral setae, Fig. 83). Sharp’s (1903) Division 2, excluding the genus Colpocaccus, is cladistically supported by synapomorphous states of characters 98 (posterior pronotal seta absent, Figs 71, 72), 154 (elytral humeri moderately narrow, Fig. 73), 156 (humeral angle tightly rounded) 201 and 203 (metathoracic flight wings vestigial, with apex just visibly extending beyond metanotum). Lacking a cladogram, the latter four characters might be interpreted as a convergent syndrome associated with brachyptery. However, based on all most parsimonious cladograms of this study exemplified in the strict consensus (Fig. 88), these characters are synapomorphous at the base of Division 2. Moreover, narrow humeri are not always associated with brachyptery, as the abacoid body form (character 153) with its broad humeri (e.g. Fig. 6) evolved at node 320 (Appendix 2), well after the origin of the brachypterous Division 2. Biogeographic analysis A fully resolved cladogram is required by COMPONENT (Page, 1993); if such is not provided, the program arbitrarily resolves polytomies. Therefore, the first of the 14 169 equally most parsimonious cladograms stored in memory was used as the basis for the biogeographic analysis. This taxon cladogram (Figs 89, 90) was converted to a taxon-area cladogram by substituting the areas of endemism for the species names. In this conversion, the natural distribution of the fourth widespread species listed above, Colpodiscus lucipetens, was assumed not to include Oahu, as the only record from that island was from a synanthropic situation. Also, monophyletic sets of areas were collapsed to a single terminal. One trichotomy remained in the taxon-area cladogram, the three clades with Metromenus meticulosus, Atelothrus metromenoides and Metromenus epicurus basal (Fig. 90). As the node subtending these clades would be optimized as Oahu regardless of resolution, the metromenoides clade was arbitrarily considered sister to the epicurus clade. The heuristic search for the area cladogram that best fit the observed taxon-area cladogram used the following options: (1) © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini 159 Table 1. Fit values for area cladograms found through heuristic tree mapping onto the resolved taxon-area cladogram (Figs 89, 90); trees A– D of Figs 91–94. Table 2. Fit values when taxon-area cladograms of the revised Division 1 and Division 2 are compared to hypotheses of area relationships based on taxon-area cladogram of entire platynine radiation. Tree Duplications Leaves added Losses Tree A B C D 60*** 60*** 60*** 60*** 244*** 257** 241*** 253*** 140*** 140*** 147** 140*** Revised Division 1, Node 310 (Fig. 87) A 22 95 B 22 100 C 25 104 D 25 112 Division 2, Node 332 (Fig. 88) A 35 140 B 35 148 C 32 128 D 32 132 ** 5 cladogram fit value less than that of 99% of randomized trees; *** fit value less than 99.9% of randomized trees. absence of a clade in an areas was assumed missing data, (2) widespread species were mapped based on relationships of endemic taxa, (3) subtree pruning and regrafting was performed and (4) numbers of duplications was the criterion minimized. A total of 612 rearrangements of the area cladogram or user tree was tried, resulting in discovery of four area cladograms requiring sixty duplications (Figs 91–94, Table 1). These four cladograms agree on (1) earliest divergence of Kauai, (2) second divergence of Oahu and (3) later divergence of the islands that comprised the superisland Maui Nui from Big Island, or Hawaii. Hypothesized relationships of areas within Maui Nui are more ambiguous. If number of added leaves (or half the number of items of error) is minimized (Fig. 93, tree C), Maui Nui vicariance is proposed to first involve divergence of Lanai, then isolation of West Maui, and finally vicariance of Haleakala (East Maui) and Molokai. This hypothesis is not supportable given the topological relationships of mountains comprising Maui Nui, chronological history of subsidence of Maui Nui (Clague & Dalrymple, 1989), nor rise of sea levels in the Pleistocene (McDonald et al., 1983). One of the three trees minimizing losses of lineages, and the second best fitting area cladogram based on minimizing items of error (Fig. 91, tree A), is more defensible given these criteria. Via this scenario, the leeward island Lanai was again first divergent, leaving the three large windward volcanoes as one primordial area. Subsequently, the largest, newest volcano Haleakala vicariated from West Maui 1 Molokai, with the latter two islands exhibiting the greatest affinities. This scenario does not agree with likely inundation events during the subsidence of Maui Nui (Clague & Dalrymple, 1989), as Molokai is separated from West Maui by the 16 km wide Pailolo Channel, whereas West Maui and Haleakala remain connected by a low isthmus. Acceptance of this vicariance hypothesis carries the attendant interpretation that vicariance was caused by factors other than inundation of sea-level habitats. Two other area cladograms require the minimum sixty duplications on the reconciled tree (Figs 92, 94, trees B, D) however, these trees require 257 and 253 added leaves, respectively, from nine to sixteen more than the two trees mentioned above. Nonetheless, both can account for their added leaves using only 140 losses, the same as tree A (Fig. 91). Examination of the taxon-area cladogram (Figs 89, 90) illustrates the conflicting area relationships that produce ambiguity in tree mapping among the Maui Nui areas, with either East Maui and Molokai, West Maui and Molokai, East © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Duplications Leaves added Losses 50 48 66 63 89 91 80 76 and West Maui, or Molokai and Lanai placed in sister area status relative to at least one other area formerly comprising Maui Nui. If the fit for the four area cladograms with minimal duplications (Table 1) is compared to the fit of 1000 randomized area cladograms generated using COMPONENT, the numbers of items of error and losses are significantly less than those for 99–99.9% of the randomized trees, supporting the nonrandomness of these taxon-area cladograms. Among randomized trees, the lowest 5% of trees have less than or equal to sixty-five duplications, 298 leaves added, and 165 losses. Given that values for leaves added and losses for tree A, and tree D (Figs 91, 94) are lower than those for 99.9% of the randomized area cladograms, one might be tempted to give preference to these over trees B and C (Figs 92, 93). However, support for a criterion of non-randomness does not directly lead to the utility of such a criterion for choosing one nonrandom tree over another. Thus, it is not possible to rigorously develop a concise hypothesis of area relationships based on results of this randomization test (Carpenter, 1992). We approached the choice of a preferred hypothesis of area relationships by another manner, i.e. a comparative analysis of support for hypotheses of relationship determined by the redefined Division 1 vs. Division 2. As the two Divisions are sister taxa, direct comparison of divergence patterns since the origin of each is possible. When the taxon-area cladograms for each Division are fit to the area cladograms resulting from analysis of the entire radiation, the area relationships defined by Division 1 require seven to thirteen fewer duplications, sixteen to fifty-three fewer added leaves, and ten to forty-three fewer losses than those defined by Division 2 (Table 2). Thus, the character ambiguity exhibited by lower support values in Division 2 (Fig. 88) is associated with poorer fit to the biogeographic pattern determined by all species. Given the better support for characters in Division 1 (Figs 89, 90), and the better fit of the area cladograms defined by the taxon-area cladogram of the entire platynine radiation to the Division 1 taxon-area cladogram, we feel that those trees that best fit the Division 1 taxon-area cladogram should be preferred, i.e. trees A and B (Table 2, Figs 91, 92). Among these, tree A (Fig. 91) requires thirteen fewer added leaves and an equal 160 James K. Liebherr and Elwood C. Zimmerman number of losses compared to tree B (Table 1), leading us to chose tree A (Fig. 91) as the preferred hypothesis of area relationships. One might argue that use of only one of the 14 169 resolved taxon cladograms as the basis for the biogeographic analysis unsuitably constrains the results. However, examination of the strict consensus cladogram (Figs 87–88) shows that component hierarchies congruent with the general area relationship of (Kauai(Oahu(Maui Nui 1 Hawaii))) are repeated twice on the cladogram (nodes 220 and 276), and Maui Nui areas are grouped together or with the island of Hawaii in numerous components (nodes 242, 249, 262, 266, 288, 304 and 315). No other single area relationship for these four major areas is prevalent. Moreover, the basal position of Kauai relative to the other areas is well established in the strict consensus. Finally, the cladogram we used (Figs 89, 90) incorporated the taxon relationship at node 218 (Fig. 87) of (Derobroscus micans (Brosconymus optatus 1 nsp. ‘hihia’)). All trees exhibiting this three-taxon statement resolved the taxa subtended by node 291 (Fig. 87) as in the taxon-area cladogram (Fig. 89). The preference for this particular three-taxon statement is based on geographical data: D. micans is widespread across Oahu, B. optatus is a Koolau Range endemic, and nsp. ‘hihia’ is a Waianae Range endemic. Given an allopatric interpretation of speciation, this arrangement is the only one among those present in the multiple equally parsimonious cladograms that is acceptable. Discussion Monophyly The monophyly of the Hawaiian Platynini is not surprising, given the exceedingly disharmonic biota of the islands. Zimmerman (1948) estimated that the currently known 5000 native insect species (Nishida, 1992) are derived from 200 to 250 ancestors. Among the Carabidae, native species are restricted to three tribes, the Bembidiini, Psydrini and Platynini (Sharp, 1903). Thus, demonstrable evidence illustrates the difficulty of Hawaiian colonization before the advent of mankind’s aid. Platynine monophyly is, conversely, extremely surprising when viewed in light of the anatomical diversity of the constituent species (Figs 1–6). Seven characters support monophyly, including five setational characters, the shape of the pronotal hind angles and the punctation of the elytral striae. All of these are reversed in the ingroup, and derived in parallel in at least one of the outgroup taxa. Given this ambiguity in character state distribution, coupled with the substantial anatomical diversity exhibited by Sharp’s genera, only the cladistic hypothesis permits us to assess the monophyly of this entire radiation. We suggest that future efforts in deciphering the phylogenetic history of Hawaiian groups may rest on similarly homoplastic character evidence, supporting cladistic analysis with all ingroup species plus many outgroup species represented as the only reliable means to identify monophyly of Hawaiian species swarms. Colonization events, even on isolated island systems, happen within the lifespan of a single individual or small group of individuals. Therefore, we should not expect extreme divergence of the basal species of an island radiation from source faunas unless extensive extinction has occurred on the islands or in the source areas. The rigorous search for nearest outgroups may shed light on the age of origin of island groups, and thereby provide an estimate for the amount of extinction that might have occurred. For example, the repeated attempts to use various single species or limited sets of species as the closest outgroup to Hawaiian drosophilid flies did not produce a robust hypothesis of the time of origin for this very diverse group (Throckmorton, 1975). Only comprehensive cladistic analysis using many exemplars has shown that the subgenus Drosophila Fallén is likely to be the outgroup to a monophyletic assemblage of Hawaiian fruit flies (DeSalle, 1992, 1995; DeSalle & Grimaldi, 1992). Evidence from the discovery of fossil Drosophila (Grimaldi, 1987), coupled with the sistergroup relationship between subgenus Drosophila and the Hawaiian forms, permits unambiguous dating of the origin of this species swarm as older than 30 million years (Ma). In this case, the extensive extinction of all taxa endemic to the high islands older than Kauai should be likely to cause trouble in making homology statements between and among the Hawaiian and mainland forms. The disparity among morphological and molecularly based estimates of phylogeny in this group (Grimaldi, 1990; DeSalle & Grimaldi, 1991) may be caused by the loss of information through extinction of annectant forms. For the Hawaiian Platynini, monophyly should be tested through the search for new character systems, and through study of the stability of the placement of the radiation’s root in a taxonomically more comprehensive array of platynine generic-level taxa. For example, comparison of the relationships advanced here with those based on characters of first-instar larvae (Liebherr, 1995) illustrates incongruence among adult and larval character data. Synapomorphies exhibited by larvae from a set of twenty-three platynine taxa of both sister subtribes Platyni and Sphodri (Liebherr, 1986) placed Platynus magnus closest to the Hawaiian Disenochus fractus. Both exhibited a derived keel-like egg burster, the sclerotized ‘egg teeth’ on the frons of first-instar carabid larvae. Whereas preliminary examination of reared larvae of Metromenus fossipennis and Colpocaccus posticatus (Liebherr, unpublished data) indicate that this state is synapomorphous throughout a broad range of Hawaiian platynines, the relationships suggested in this study, with P. magnus placed well away from the Hawaiian forms, support parallel derivation of keel-like egg bursters in the Hawaiian radiation and P. magnus. Larval rearing from a diversity of australasian platynine taxa is required to permit fruitful discussion of this character. Character evolution The most striking difference in the overall pattern of character evolution in this radiation is the relatively strong character support, resulting in a resolved hypothesis of relationships within Division 1 (support value average of 5.5 steps per node), vs. the very poor and conflicting character support existing within © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini Division 2 (support value average of 2.7). This conclusion is made even stronger by the fact that the strict consensus upon which we are measuring character support (Figs 87, 88) is less resolved for Division 2, necessitating, if anything, the addition of more steps to character support values at polytomous nodes because of required parallel character transformations in all unresolved edges attaching to such nodes. We may also view this disparity in character evolution by comparing the total number of character state transformations in the two sister Divisions, discernible by summing the horizontal branch lengths of a fully resolved cladogram (e.g. Figs 89, 90). The total number of steps in Division 1 from its common ancestor with Division 2 is 602 steps; the converse is 448 steps. The 34% greater total path length within Division 1 exists in the face of 32% fewer species; fifty Division 1 species vs. seventy-three in Division 2. This disparity in character transformation associated with cladistic divergence may be based on intrinsic attributes of the sister lineages, whereby Division 1 taxa have undergone greater changes in cuticular evolution due to underlying genetic or developmental mechanisms. Conversely, if character change is preserved by speciation (Futuyma, 1989), the greater divergence across Division 1 could be explainable through a greater speciation rate, coupled with a higher subsequent extinction rate for that lineage. Thus, greater character evolution would be based on history. Our biogeographic analysis identified the hypotheses of area relationships that minimized the items of error added to the taxonarea cladogram for the entire platynine radiation. If intrinsic mechanisms are causing greater character evolution in Division 1, the number of losses of lineages should be similar between the two lineages, i.e. the fit of the taxon-area cladogram for each Division to the fundamental area cladogram based on all taxa should be the same. Conversely, if a greater speciation rate coupled with greater subsequent extinction explains the greater character divergence but fewer species in Division 1, we should see a better fit for the Division 2 vs. Division 1 cladograms. As reported above, we found that relationships within Division 1 actually exhibited a better fit to the overall pattern of area relationships, seriously compromising any attempt to imply greater speciation rate with associated extinction in Division 1. Therefore, it appears that underlying differences in cuticular evolution better explain the difference in character support between the two Divisions. It is beyond our scope to present a detailed discussion of the evolutionary patterns in all character systems in this first report. Nonetheless, it is of interest to examine the patterns implied by our cladistic hypothesis for three of the character systems used by Sharp (1903) in the only previous classification, i.e. flightwing configuration, tarsal structure and pronotal setation. Evolution of brachyptery. We did not utilize any ad-hoc weighting scheme for flight-wing configuration. Of the 206 characters, wing reduction accounted for five, with one character dealing with the peculiar strap-like remnant of wings observable in Bryanites samoaensis and nsp. ‘apterum.’ The Hawaiian species were coded in a straightforward manner into five states, four increasingly derived: (1) wing ratio 3.0–5.03 (plesiomorphic), (2) wing ratio 1.4–2.3 (character 200), (3) wing © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 161 ratio 0.4–1.1 (character 201), (4) wing remnant extending beyond metanotum (character 203) and (5) wing remnant not extending beyond metanotum (character 204). We did not code the shortening of the metathorax observed in concert with the evolution of brachyptery. Changes in body shape in some brachypterous species, associated with dorsal convexity of the elytra and a tubular body, often caused narrowing of the metepisternum, the most obvious sclerite associated with metathoracic reduction. Thus metepisternal shape, or length to width ratios, were not informative, being confounded by wing loss and body shape evolution. As species did not exhibit variation in wing configuration, a confirmation of Sharp (1903), we interpret this sequence as a straightforward pattern of diminution through release from the necessity to use the wings for dispersal. Southwood’s (1977) criteria of isolation and stability as factors releasing taxa from selective pressures requiring dispersal are eminently exemplified, at least for Carabidae, by Hawaiian montane ecosystems. Southwood’s third criterion, favourableness, which might be called into account in the evolution of brachyptery, is not so much in evidence, as the predominantly montane Platynini must deal with cool temperatures, extremely wet conditions, and the scarcity of food resources when individuals are isolated on small patches of habitat due to flooding. Nonetheless, such conditions are not suitable for beetle flight anyway, suggesting that advantages to dispersal may be obviated by the environmental conditions facing taxa throughout much of their existence. Of the 128 species, thirteen possess flight wings with either a wing ratio of 3.0–5.0, or 1.4–2.3. These species are: (1) nsp. ‘mandibularis’, (2) the four Colpocaccus species, (3) Baryneus sharpi, (4) Colpodiscus lucipetens, and nsp. ‘kukui’, (5) Mysticomenus tibialis and Mysticomenus mysticus and (6) the three species of Chalcomenus Sharp (Fig. 87). All species cluster near the root of the ingroup, and no reversions from a brachypterous condition to a more fully winged condition are mandated by the cladogram. The reduction of the alar surface from the plesiomorphic state to a wing ratio of 1.4–2.3 may be interpreted as three forward steps: (1) at node 277, (2) in Apteromesus maculatus and (3) at node 309. A less pleasing alternative is the reduction of wings at node 333, and reacquisition of long wings: (1) in Baryneus sharpi, (2) at node 279 and (3) at node 239 (Fig. 87). The latter interpretation assumes that a functional flight apparatus could be reinstated after a period of evolutionary absence. In this case, that would involve not only the reinstatement of wings three times, but also three instances of reversal of an abbreviated metathorax to the full-sized condition. As both optimizations are based on equally parsimonious character evidence, there is no need to reject the interpretation of unidirectional reduction of flight wings. Tarsal structure. In the plesiomorphic tarsal configurations represented in this data set, the basal tarsomeres exhibit dorsolateral sulci. These are best developed in the mid- and hindlegs (Figs 76, 77). Given that Sharp used this criterion to diagnose his Division 2, it is not surprising that this assemblage is paraphyletic under cladistic analysis. The removal of Colpocaccus to the position of basal sister group to Division 1, plus a redefined Division 2, is warranted. The reduction of 162 James K. Liebherr and Elwood C. Zimmerman dorsolateral sulci is associated with a robust body and heavy legs. Thus, the sulci are interpretable as strengtheners of slender tarsi. Taxa with robust legs can accomplish the same function through heavily sclerotized, more tubular tarsomeres, e.g. Division 1 Hawaiian platynines and many Pterostichini, a related carabid tribe. The ventral vestiture of the tarsomeres is correlated with preferred substrate. The most highly developed tarsal pads on tarsomeres 3 and 4 of Colpodiscus (Fig. 85) are replete with simple and straplike climbing setae (Stork, 1980). Individuals of this species have been collected in flight at lights, and on plant surfaces. Other arboreal species, such as Atelothrus dyscoleus of East Maui, most commonly beaten from plants such as ieie (Freycinetia arborea Gaudichaud-Beaupré), exhibit narrower fourth tarsomeres with only simple adhesive setae (Fig. 84). Nonetheless, the ventrolateral bands of setae on tarsomeres 1–4 are broader and thicker than in those of more strictly terrestrial species, such as the Kauaian Mesothriscus alternans (Fig. 81). The ventral setation of the fifth tarsomere also varies substantially, from short pegs not visible except under high magnification (e.g. M. alternans, Fig. 81), to longer, clearly visible setae (Metromenus caliginosus, Fig. 82), to long and numerous setae closely spaced in rows (Barypristus rupicola, Fig. 83). Extremely long setae on the fifth tarsomere occur only rarely, and in all cases in terrestrial species: (1) the two Barypristus plus Baryneus sharpi (Fig. 87, node 278), (2) the cave-inhabiting Atelothrus aaae (Liebherr & Samuelson, 1992) and (3) Mauna frigida from Haleakala. The initial scoring of two types of elongation in the fourth tarsomere apical lobes is justified by the resultant cladogram, as derivations in characters 193–194, i.e. the equal elongation of inner and outer lobes observed in the Hawaiian species, and character 195, i.e. the elongation of only the outer apical lobe observed in outgroup lineages, are independently derived from more plesiomorphic conditions. Both Notagonum submetallicum and the Lorostema species patristically close to the Hawaiian radiation exhibit nonlobate fourth tarsomeres. Outer lobe elongation (character 195) occurs at nodes 188 and 204 in the outgroups (Fig. 86). Conversely, substantial elongation of the fourth tarsomere lobes in the Hawaiian species supports node 280 (character 193) and node 279 (character 194) (Appendix 2), situated well within Division 1 (Fig. 87). Pronotal configuration. Sharp (1903) used pronotal setation extensively in his generic diagnoses. It was this character system under which he admitted that an individual could ’pass from the genus of its parents to another’ (Sharp, 1903). Indeed, the setation of the pronotal margins, so easily viewed and useful for quick identification, is not a good indicator of affinity. A quick scan of the cladogram for Division 2 (Fig. 88) shows polyphyly of Atelothrus Sharp (possessing the basal pronotal seta, e.g. Fig. 70), Mesothriscus Sharp (possessing the lateral pronotal seta, Fig. 71), and Metromenus Sharp (lacking either setae, Fig. 72). The underlying developmental basis for these setae must be extremely plastic, for the characters they imply change numerous times (sixteen times for the lateral seta, twenty times for the basal seta). Moreover, ten species are polymorphic for presence of the lateral setae, and eleven are polymorphic for presence of the basal setae (Appendix 1). Although they are obvious characters, their use in the diagnosis of any higher groups within the Hawaiian platynini must be eschewed. Just as homoplasy is ensured by the independent coding of pronotal setation in company with all possible combinations of lateral and basal pronotal setal presence and absence (Figs 57, 70–72), the breadth of the pronotal lateral margin, and whether it is beaded or not (characters 93–94) necessarily require homplasy due to observation of all four possible state combinations. As an example of how this coding can uncover different patterns of character evolution, we can look at the distribution of these two characters in different clades of Division 2 (Fig. 88). In one clade subtended by Platynus ambiens, character 93 supports node 323 and reverses to the plesiomorphic state at node 318. Character 94 represents a subsequent more restricted derivation, supporting nodes 164, 301 and 134 (Fig. 88). Among these taxa, a moderately broad, beaded pronotal margin transforms by narrowing while retaining the marginal bead. Looking instead at the clade rooted at the polytomous node 331, character 94 is more broadly distributed, supporting nodes 226 and 272. Conversely, character 93 arises independently in Metromenus meticulosus, nsp. ‘foveolata’, and at node 271. In these cases we observe a narrow margin without a bead transforming to a narrow, beaded margin. Reticulate coding permits these alternate pathways, recognizing that developmental transformations may vary even among a geographically isolated monophyletic radiation. Bursal asymmetry. Because they are sister species (Fig. 87, node 277), the mirror-like asymmetrical difference between the bursa of Barypristus rupicola and B. incendiarius (Figs 35, 36) has occurred at or after the speciation event that isolated these species. The plesiomorphic ‘left-handed’ bursa of B. incendiarius has transformed to a ‘right-handed’ structure in B. rupicola. Inversion of bursal asymmetry has likewise occurred in Prodisenochus terebratus relative to the more plesiomorphic ‘left-handed’ bursal expansion observed in its adelphotaxon, Broscomimus lentus (Fig. 87, node 302) and cladistically more basal relatives. Proper dorsal-ventral polarity of the Drosophila compound eye is controlled by the mirror gene, a homeotic gene that governs development of the equator of the eye, that region demarcating mirror-image fields of opposing forms of ommatidia (McNeill et al., 1997). Left-right symmetry control may also be governed by such a homeotic gene, permitting mutations to change symmetry in a single event. Such genes may also be responsible for aedeagal inversion, i.e. left-right switches in aedeagal symmetry, in the sphodrine carabid genera Pristosia Motschulsky (Lindroth, 1956) and Calathus Bonelli (Ball & Negre, 1972). Switched aedagal asymmetry is not present in males of B. rupicola or P. terebratus. Age of origin and source Kauai represents the root of the platynine radiation among the extant high islands. In this sense, the biogeographic pattern of the Platynini is progressive. However, unlike the hypothetical progression rule patterns advanced by Funk & Wagner (1995; Figs 17.6–17.7), the platynine pattern combines degrees of vicariance of Maui Nui, with the Big Island the sister area to all the islands comprising the former Maui Nui, and newer Big © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini Island–Maui connections in some cases indicated by widespread species (Figs 89, 90). We prefer to use the fundamental area cladogram (Figs 91–94) to identify the overall pattern. Based on this, the platynine pattern can be called progressive vicariance. It has operated on Oahu as well as Maui Nui, as on the former there are thirty-two species, of which eight are Waianae Range endemics, fourteen are Koolau Range endemics, and only ten are widespread across the island. Placement of Kauai basally on all possible fundamental area cladograms (Figs 91–94) implies that the Hawaiian Platynini are at least as old as that island, i.e. 5.1 Ma (Clague & Dalrymple, 1989). The numerous occurrences of Kauai on the taxon-area cladogram (Figs 87, 88) suggests representation of numerous independent lineages on Kauai, consistent with prior radiation on now drowned or highly eroded north-eastern islands. The history of island formation in the Hawaiian chain constrains the oldest likely source as Kure, aged 29 Ma. No islands higher than 1000 m elevation existed between Kure and Koko, aged 48 Ma (Carson & Dalrymple, 1995). Koko and all of these smaller intermediate-aged volcanoes had subsided before Kure became subaerial, suggesting that continuous within-archipelago colonization has only been possible for the past 29 Ma. Such a window of within-archipelago colonization, 5.1–29.0 Ma, is roughly consistent with the hypothesis placing the time of origin of Hawaiian Drosophilidae as ù 30 Ma (DeSalle, 1995). A more restrictive hypothesis, whereby Kauai was the source of colonization by a single propagule, would be better supported if most occurrences of Kauai grouped near the base of the cladogram, indicating a single lineage that subsequently spawned propagules that colonized Oahu and subsequent islands. Arguing extinction patterns from present-day patterns is fraught with assumptions, and independent data such as that seen in the Drosophilidae are required to convincingly answer the question of time of origin. The closest outgroup to the native Hawaiian species is the genus Lorostema which, in addition to L. informalis from New Guinea and L. ogurae from Japan (Fig. 86), includes L. bothriophora Redtenbacher from Queensland, Australia, Tahiti (Moore et al., 1987) and Rapa, where it has been collected at light (Britton, 1938), the type species, L. alutaceum Motschulsky from India, Sri Lanka and Burma, and L. subnitens Andrewes from Sumatra (Csiki, 1931). That all of these species are known from islands, although Sri Lanka is a continental island, suggests substantial colonization ability. In this, Lorostema species exhibit an attribute necessary for the common ancestor of the Hawaiian platynine radiation. Lorostema is placed patristically adjacent to the root of the cladogram, which was set by choosing Notagonum submetallicum as the outgroup for the entire analysis. Whether N. submetallicum would remain the root of this complex in an analysis including many more representatives of the µ 350 platynine genera remains to be seen. Our choice of this taxon as the outgroup was based on an overall estimate of its similarity to a platynine ground-plan. In order to verify the position of the Hawaiian radiation within Platynini as a whole, the 128 species of the Hawaiian radiation should be reduced to a single term by coding them with the states at the base (node 335) of the ingroup, and then conducting an analysis with as many generic © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 163 representatives across the Platynini as possible, including genera in the subtribe Sphodri (Liebherr, 1986) and outgroups to the Platynini. Biogeographic relationships among various outgroups may then prove useful for dating nodes near the base of the ingroup, providing a minimum estimate for the age of this radiation beyond the 5.1 million year lifespan of Kauai. The hypothesis that Lorostema is the sister group to the Hawaiian Platynini places the geographical source of the dispersal resulting in the platynine radiation as Asian or Australian. In this, the Platynini agree with the dispersal pattern attributed to the other major Hawaiian carabid radiation; the 168 known Hawaiian species (unpublished data) of the genus Mecyclothorax Sharp (Britton, 1948). In addition to Hawaii, Mecyclothorax is represented by species in Java, New Guinea, Australia, New Zealand, New Caledonia and several south Indian Ocean islands as well as an extensive radiation on Tahiti (Perrault, 1978, 1979, 1984, 1986, 1987, 1988, 1989, 1992). We know that dispersal events leading to founding of the two lineages in Hawaii were not contemporaneous, for the Hawaiian Mecyclothorax are absent from Kauai, and the most plesiomorphic taxa reside on Maui Nui (Britton, 1948), supporting a founding event for this radiation no earlier than the origin of Molokai, i.e. 1.9 Ma ago (Clague & Dalrymple, 1989). An obvious question to ask is whether area relationships for Mecyclothorax from Maui Nui are congruent with, i.e. allow unequivocal support for, a particular pattern supported by the Platynini. Only comprehensive cladistic analysis as presented above will allow an answer to that question. The primary goal of this study was to identify phylogenetic relationships within the Hawaiian Platynini, and to determine whether the species swarm is monophyletic. This has been accomplished, and we conclude that the complex is indeed derived from a single colonizing taxon. The finding of monophyly offers the opportunity to consider the many biological attributes of this radiation as homologous modifications from those of a common ancestor. Parallel occurrences of such transformations across the phylogeny of the group, although homoplastic, permit evaluation of the interplay of ecological and historical factors during the evolution of this species swarm. We see three major avenues upon which this study should be advanced: (1) the incorporation of new characters to test aspects of the phylogenetic hypothesis for the Hawaiian platynine species, (2) more comprehensive study of platynine generic-level taxa to better determine the historical context of the Hawaiian Platynini and (3) study of other Hawaiian taxa, especially those most diverse in areas of endemism comprising the former Maui Nui, to determine the generality of area relationships in this vicariant system. Acknowledgements This project has taken a number of years to complete, having its origins in E.C.Z.’s studies in Hawaii during the 1940s and at the British Museum of Natural History in the 1950s. We must therefore thank many people for assisting in its completion. All of the taxonomic material and types were studied by E.C.Z. and 164 James K. Liebherr and Elwood C. Zimmerman J.K.L. in London, over a 40-year span, with necessary loans graciously made by N. E. Stork and S. J. Hine of The Natural History Museum. G. A. Samuelson and G. M. Nishida of the Bernice P. Bishop Museum, Honolulu, hosted J.K.L. during his work there. Extensive new material has been collected over the period of our studies. We thank A. Asquith, E. J. Ford, A. E. Hajek, F. G. Howarth, A. C. Medeiros, D. A. Polhemus and J. M. Valentine for their collecting efforts. The laboratory research greatly benefitted from the dedication of graduate research assistants including M. J. McDonald, C. J. Marshall, J. L. Santisteban and K. W. Will. Teresa Wells and Zack Falin assisted in preparation of field material. Curtis P. Ewing produced the inked illustrations. Barry P. Moore, CSIRO, Canberra, shared information about Fijian platynines. We thank Scott Miller, Tina Kuklenski and Dan Polhemus for their ‘aloha’ spirit and commitment to helping J.K.L. feel at home in Hawaii. We also thank Rod Page for critical review of the manuscript. The National Geographic Society (Grant no. 4431-90) and the National Science Foundation (Grant no. DEB-9208269) provided support to J.K.L. This is contribution no. 1996-009 of the Hawaii Biological Survey, and it incorporates data from E.C.Z.’s unpublished manuscript for his proposed Volume 16 of Insects of Hawaii. References Asquith, A. (1994) An unparsimonious origin for the Hawaiian Metrargini (Heteroptera: Lygaeidae). Annals of the Entomological Society of America, 87, 207–213. Ball, G.E. & Negre, J. (1972) The taxonomy of the Nearctic species of the genus Calathus Bonelli (Coleoptera: Carabidae: Agonini). Transactions of the American Entomological Society, 98, 412–533. Blackburn, T. (1877) Characters of a new genus, and descriptions of new species, of Geodephaga from the Sandwich Islands. The Entomologists’ Monthly Magazine, 14, 142–148. Blackburn, T. (1878) Characters of new genera and descriptions of new species of Geodephaga from the Hawaiian Islands. The Entomologists’ Monthly Magazine, 15, 119–123, 156–158. Blackburn, T. (1879) Characters of new genera and descriptions of new species of Geodephaga from the Hawaiian Islands. The Entomologists’ Monthly Magazine, 16, 104–109. Blackburn, T. (1881) Characters of new genera and descriptions of new species of Geodephaga from the Hawaiian Islands. The Entomologists’ Monthly Magazine, 17, 226–229. Blackburn, T. (1884) Notes on some Hawaiian Carabidae. The Entomologists’ Monthly Magazine, 21, 25–26. Bremer, K. (1988) The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution, 42, 795–803. Britton, E.B. (1938) Carabidae of the Society Islands and Rapa (Coleoptera). Occasional Papers of the Bernice P. Bishop Museum, 14, 103–110. Britton, E.B. (1948) A revision of the Hawaiian species of Mecyclothorax (Coleoptera: Carabidae). Occasional Papers of the Bernice P. Bishop Museum, 19, 107–166. Carlquist, S. (1965) Island Life. Natural History Press, Garden City, New York. Carpenter, J.M. (1992) Random cladistics. Cladistics, 8, 147 – 153. Carson, H.L. & Dalrymple, G.B. (1995) Geology and biogeography of the Hawaiian Islands. Hawaiian Biogeography (ed. by W. L. Wagner and V. A. Funk), pp. 14–29. Smithsonian Institution Press, Washington, D.C. Clague, D.A. & Dalrymple, G.B. (1989) The Hawaiian-Emperor Chain. The Geology of North America, N (The Eastern Pacific Ocean and Hawaii) (ed. by E. L. Winterer, D. M. Hussong and R. W. Decker), pp. 187–287. Geological Society of America, Boulder, Colorado. Csiki, E. (1931) Coleopterorum Catalogus 115 (Carabidae: Harpalinae V.), pp. 739–1022. W. Junk, Berlin. Darlington, P.J., Jr (1952) The carabid beetles of New Guinea. 2. The Agonini. Bulletin of the Museum of Comparative Zoology, 107, 89– 252 1 4 pls. Darlington, P.J., Jr (1971) The carabid beetles of New Guinea. IV. General considerations; analysis and history of taxa; taxonomic supplement. Bulletin of the Museum of Comparative Zoology, 142, 129–337. DeSalle, R. (1992) The origin and possible time of divergence of the Hawaiian Drosophilidae: evidence from DNA sequences. Molecular Biology and Evolution, 9, 905–916. DeSalle, R. (1995) Molecular approaches to biogeographic analysis of Hawaiian Drosophilidae. Hawaiian Biogeography (ed. by W. L. Wagner and V. A. Funk), pp. 72–89. Smithsonian Institution Press, Washington, D.C. DeSalle, R. & Grimaldi, D.A. (1991) Morphological and molecular systematics of the Drosophilidae. Annual Review of Ecology and Systematics, 22, 447–475. DeSalle, R. & Grimaldi, D. (1992) Characters and the systematics of Drosophilidae. Journal of Heredity, 83, 182–188. DeSalle, R. & Grimaldi, D. (1993) Phylogenetic pattern and developmental process in Drosophila. Systematic Biology, 42, 458– 475. Farris, J.S. (1970) Methods for computing Wagner trees. Systematic Zoology, 19, 83–92. Farris, J.S., Kluge, A.G. & Eckhardt, M.J. (1970) A numerical approach to phylogenetic systematics. Systematic Zoology, 19, 172–191. Funk, V.A. & Wagner, W.L. (1995) Biogeographic patterns in the Hawaiian Islands. Hawaiian Biogeography (ed. by W. L. Wagner and V. A. Funk), pp. 379–419. Smithsonian Institution Press, Washington, D.C. Futuyma, D.J. (1989) Speciational trends and the role of species in macroevolution. American Naturalist, 134, 318–321. Gillespie, R.G., Croom, H.B. & Palumbi, S.R. (1994) Multiple origins of a spider radiation in Hawaii. Proceedings of the National Academy of Sciences, U.S.A., 91, 2290–2294. Goloboff, P.A. (1995) NONA Version 1.5 [32 bit] [available from the author at Fundacion Miguel Lillo, 205, 4000 San Miguel de Tucuman, Argentina]. Gressitt, J.L. (1978) Evolution of the endemic Hawaiian cerambycid beetles. Pacific Insects, 18, 137–167. Grimaldi, D.A. (1987) Amber fossil Drosophilidae (Diptera) with particular reference to the Hispaniolan taxa. American Museum Novitates, 2880, 1–23. Grimaldi, D.A. (1990) A phylogenetic, revised classification of genera in the Drosophilidae (Diptera). Bulletin of the American Museum of Natural History, 197, 1–139. Kaneshiro, K.Y., Gillespie, R.G. & Carson, H.L. (1995) Chromosomes and male genitalia of Hawaiian Drosophila. Hawaiian Biogeography (ed. by W. L. Wagner and V. A. Funk), pp. 57–71. Smithsonian Institution Press, Washington, D.C. Liebherr, J.K. (1986) Cladistic analysis of north american Platynini and revision of the Agonum extensicolle species group (Coleoptera: Carabidae). University of California Publications in Entomology, 106, 1–198. Liebherr, J.K. (1988) Biogeographic patterns of West Indian Platynus beetles. Zoogeography of Caribbean Insects (ed. by J. K. Liebherr), pp. 121–152. Cornell University Press, Ithaca, New York. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini Liebherr, J.K. (1992) Phylogeny and revision of the Platynus degallieri species group (Coleoptera: Carabidae: Platynini). Bulletin of the American Museum of Natural History, 214, 1–115. Liebherr, J.K. (1995) Description of the larval stages of Disenochus fractus Sharp and the origin of the Hawaiian platynine Carabidae (Coleoptera). Systematic Entomology, 20, 17–26. Liebherr, J.K. (1998) On Rembus (Colpodes) brunneus MacLeay (Coleoptera: Carabidae, Platynini): redescription and relationships. Journal of Natural History, 32, in press. Liebherr, J.K. & Polhemus, D.A. (1997) R. C. L. Perkins: 100 Years of Hawaiian Entomology. Pacific Science, 51, 343–355. Liebherr, J.K. & Samuelson, G.A. (1992) The first endemic troglobitic carabid beetles in Hawaiian lava tubes (Coleoptera: Carabidae). PanPacific Entomologist, 68, 157–168. Liebherr, J.K. & Will, K.W. (1996) New North American Platynus (Coleoptera: Carabidae), a key to species north of Mexico, and notes on species from the Southwestern United States. Coleopterists Bulletin, 50, 301–320. Lindroth, C.H. (1956) A revision of the genus Synuchus Gyllenhal (Coleoptera: Carabidae) in the widest sense, with notes on Pristosia Motschulsky (Eucalathus Bates) and Calathus Bonelli. Transactions of the Royal Entomological Society, London, 108, 485–574. Lindroth, C.H. (1966) The ground-beetles (Carabidae, excl. Cicindelinae) of Canada and Alaska. 4. Opuscula Entomologica Supplementum, 29, 409–648. Louwerens, C.J. (1953) The Oriental species of Colpodes Macl. (Col. Carabidae). Treubia, 22, 75–151. Maddison, D.R. (1991) The discovery and importance of multiple islands of most- parsimonious trees. Systematic Zoology, 40, 315–328. Manning, A. (1986) The Sandwich Islands Committee, Bishop Museum, and R. C. L. Perkins: cooperative zoological exploration and publication. Bishop Museum Occasional Papers, 26, 1–46. McDonald, G.A., Abbott, A.T. & Peterson, F.L. (1983) Volcanoes in the Sea: the Geology of Hawaii. University of Hawaii Press, Honolulu. McNeill, H., Yang, C.H., Brodsky, M., Ungos, J. & Simon, M.A. (1997) Mirror encodes a novel PBX-class homeoprotein that functions in the definition of the dorsal-ventral border of the Drosophila eye. Genes and Development, 11, 1073–1082. Moore, B.P., Weir, T.A. & Pyke, J.E. (1987) Zoological Catalogue of Australia. 4. (Rhysodidae and Carabidae), pp. 20–320. Bureau of Flora and Fauna, Canberra. Nelson, G. & Platnick, N. (1981) Systematics and Biogeography, Cladistics and Vicariance. Columbia University Press, New York. Nishida, G.M. (1992) Hawaiian Terrestrial Arthropod Checklist. Hawaii Biological Survey, Honolulu. Nixon, K.C. (1995) CLADOS Version 1.4.95 [32 bit] beta version (available from the author at L.H. Bailey Hortorium, Cornell University, Ithaca, NY 14853, U.S.A.). Nixon, K.C. & Carpenter, J.M. (1993) On outgroups. Cladistics, 9, 413–426. Page, R.D.M. (1990) Component analysis: a valiant failure? Cladistics, 6, 119–136. Page, R.D.M. (1991) Random dendrograms and null hypotheses in cladistic biogeography. Systematic Zoology, 40, 54–62. Page, R.D.M. (1993) COMPONENT Version 2.0. The Natural History Museum, London. Perkins, R.C.L. (1913) Introduction, being a review of the land-fauna of Hawaii. Fauna Hawaiiensis, pp. i–cccxxviii. Cambridge University Press. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 165 Perrault, G.G. (1978) La faune des Carabidae de Tahiti. II. Genre Mecyclothorax (Sharp). Nouvelle Revue d’Entomologie, 8, 27– 36 133–162. Perrault, G.G. (1979) La faune des Carabidae de Tahiti. III. Mecyclothorax fosbergi n. sp. Nouvelle Revue d’Entomologie, 9, 9–10. Perrault, G.G. (1984) La faune des Carabidae de Tahiti. VI. Révision du genre Mecyclothorax (Sharp) (Psydrini). 1. Le groupe de M. muriauxi Perrault (Coleoptera). Nouvelle Revue d’Entomologie (N.S.), 1, 19–31. Perrault, G.G. (1986) La faune des Carabidae de Tahiti. VII. Révision du genre Mecyclothorax (Sharp) (Psydrini). 2. Les groupes de M. striatopunctatus n. sp., M. dannieae Perrault, M. marginatus Perrault et M. viridis Perrault (Coleoptera). Nouvelle Revue d’Entomologie (N.S.), 3, 439–455. Perrault, G.G. (1987) Microendemisme et speciation du genre Mecyclothorax (Coleoptera – Carabidae Psydrini) à Tahiti (Polynesie Francaise). Bulletin de la Societé Zoologique de France, 112, 419–427. Perrault, G.G. (1988) La faune des Carabidae de Tahiti. VIII. Révision du genre Mecyclothorax (Sharp) (Psydrini). 3. Les groupes de M. altiusculus Britton et de M. gourvesi Perrault (Coleoptera). Nouvelle Revue d’Entomologie (N.S.), 5, 229–245. Perrault, G.G. (1989) La faune des Carabidae de Tahiti. IX. Révision du genre Mecyclothorax (Sharp) (Psydrini). 4. Le groupe de M. globosus Britton (Coleoptera). Nouvelle Revue d’Entomologie (N.S.), 6, 57–70. Perrault, G.G. (1992) Endemism and biogeography among Tahitian Mecyclothorax species (Coleoptera: Carabidae: Psydrini). The Biogeography of Ground Beetles on Mountains and Islands (ed. by G. R. Noonan, G. E. Ball and N. E. Stork), pp. 201–215. Intercept Ltd, Andover, U.K. Polhemus, D.A. (1997) Phylogenetic analysis of the Hawaiian damselfly genus Megalagrion (Odonata: Coeagrionidae): Implications for biogeography, ecology, and conservation biology. Pacific Science, 51, 395–412. Sharp, D. (1878) Description of a new species probably indicating a new genus of Anchomenidae, from the Sandwich Islands. The Entomologists’ Monthly Magazine, 14, 179–180. Sharp, D. (1884) On some genera of the sub-family Anchomenini (Platynini Horn) from the Hawaiian Islands. The Entomologists’ Monthly Magazine, 20, 217–219. Sharp, D. (1900) Letter to R. C. L. Perkins. 1 November 1900. Ms. in B. P. Bishop Museum Archives, ms. group 141, box 2. Sharp, D. (1903) Coleoptera 2. Fauna Hawaiiensis, Vol. 3, Part 3, pp. 175–292. Cambridge University Press. Southwood, T.R.E. (1977) Habitat, the templet for ecological strategies? Journal of Animal Ecology, 46, 337–365. Stork, N.E. (1980) A scanning electron microscope study of the tarsal adhesive setae in the Coleoptera. Zoological Journal of the Linnean Society, 68, 173–306. Swofford, D.L. (1991) PAUP: Phylogenetic Analysis Using Parsimony, Version 3.0. Illinois Natural History Survey, Champaign. Throckmorton, L.H. (1975) The phylogeny, ecology, and geography of Drosophila. Handbook of Genetics, 3, 421–469. Wagner, W.H., Jr (1980) Origin and philosophy of the groundplandivergence method of cladistics. Systematic Botany, 5, 173–193. Wagner, W.L., Herbst, D.R. & Sohmer, S.H. (1990) Manual of the Flowering Plants of Hawaii. Bishop Museum Special Publication 83 (2 vols). University of Hawaii Press, Honolulu. Zimmerman, E.C. (1948) Insects of Hawaii 1. University of Hawaii Press, Honolulu. Accepted 27 July 1997 166 James K. Liebherr and Elwood C. Zimmerman Appendix 1. Data matrix for 170 taxa and 206 unit-coded characters. Taxon numbers are those preceding names in Figs 86–88. Characters initially considered primitive relative to a platynine ground plan scored 0; characters initially scored derived coded 1; missing data and characters polymorphic within taxa scored ‘?’; inapplicable data scored ‘–’. © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 167 168 James K. Liebherr and Elwood C. Zimmerman © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 169 170 James K. Liebherr and Elwood C. Zimmerman Appendix 2. Unambiguously assigned character state changes for strict consensus cladogram (Figs 86–88). For internal nodes, state changes represented by the strict consensus are listed first. Those state changes defined on some but not all of the equally parsimonious trees used to construct the strict consensus are indicated by asterisk (*). © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 Hawaiian Platynini © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172 171 172 James K. Liebherr and Elwood C. Zimmerman © 1998 Blackwell Science Ltd, Systematic Entomology, 23, 137–172

RELATED PAPERS

RELATED TOPICS

Log In

Cladistic analysis, phylogeny and biogeography of the Hawaiian Platynini (Coleoptera: Carabidae

Cladistic analysis, phylogeny and biogeography of the Hawaiian Platynini (Coleoptera: Carabidae

Related Papers

RELATED PAPERS

RELATED TOPICS