Systematic Entomology (1990) 15,101-1 17
Revision of the dolichoderine ant genus Turneria (Hymenoptera: Formicidae)
STEVEN 0. SHATTUCK Department of Entomology, University of California, Davis, California, U.S.A.
ABSTRACT. The Australasian ant genus Turneria is revised. T.bidentata Forel, T.dahlii Forel and T.pacijica Mann are redescribed and a lectotype designated for T.bidentata. T.frenchi Forel is transferred to Stigrnacros (comb.n.), and T.butteli Forel is transferred to Zridornyrmex (c0mb.n.). Three new species, T.arbusta, T.collina and T.postomma, are described from Papua New Guinea. Males of the genus are described for the first time. A key to the workers of recognized species and a distribution map are provided. The group is analysed cladistically and the resulting species re- lationships are discussed, along with a comment on the use of polymorphic characters.
Introduction In this paper I redescribe three of the pre- viously known species, describe three species as The dolichoderine ant genus Turneria contains new, transfer one species to Stigmacros and one six species occurring in Papua New Guinea, to Iridomyrmex. Additionally, I give the first Australia, the Solomon Islands and Vanuatu description of the male of this genus. (Fig. 23). It is not known from New Caledonia or Fiji. All species are arboreal in parklands or Depositories primary and secondary rain forests and at least three species nest in dead twigs (Wilson, 1959, Material for this study was borrowed from the and below). following Museums (with abbreviations). Turneria was established by Forel (1895) for Australian National Insect Collection, Can- the single species T.bidentata from Australia. berra, A.C.T., Australia (ANIC); B. P. Bishop Forel subsequently added three more species: Museum, Honolulu, Hawaii, U.S.A. (BPBM); dahlii from Papua New Guinea, frenchi from British Museum (Natural History), London, Australia, and butteli from Sumatra, Indonesia. England, U.K. (BMNH); California Academy The last two species belong to Stigmacros and of Sciences, San Francisco, California, U.S.A. Iridomyrmex, respectively (see below). A fifth (CASC); Museum of Comparative Zoology, species, pacifica, was described from the Harvard University, Cambridge, Massachu- Solomon Islands by Mann (1919), who also pro- setts, U.S.A. (MCZC); Museum d’Histoire vided a key to species based on workers. The Naturelle, Geneva, Switzerland (MHNG); larvae were discussed by Wheeler & Wheeler Hope Entomological Collections, University (1974) and the proventriculus by Eisner (1957). Museum, Oxford, England, U.K. (OXUM); (The species used in these latter studies cannot P. S. Ward collection, University of California, be determined because they are based on a Davis, California. U.S.A. (PSWC); S. 0. Shat- mixed collection of dahlii and pacifica.) Wilson tuck collection, University of California, Davis, (1959, 1962) discusses the ecology of these ants California, U.S.A. (SOSC); United States and describes the queens of two species. National Museum of Natural History, Correspondence: Dr S. 0. Shattuck. Department of Washington, D.C., U.S.A. (USNM); Museum Entomology, University of California, Davis, CA fur Naturkunde der Humboldt-Universitat zu 95616, U.S.A. Berlin, Berlin, G.D.R. (ZMHB). 101 102 Steven 0.Shattuck Measurements and indices CI Cephalic index: HW/HL. EL Eye length: measured in full face view Morphological measurements were made at 50 X (Fig. 1). on a Wild stereo microscope, with a dual-axis ES Eye spread: distance between inner- stage micrometer wired to Autometronics most edges of eyes measured in full digital readouts. All measurements were re- faceview (Fig. 1). corded in thousandths of millimetres, but are EW Eye width measured in full face view expressed here to the nearest hundredth. (Fig. 1). All head measurements (LES, EW, ES, HW, FFL Fore femur length: maximum length of OOD, EL, OCD, CL, HL) were taken in full- fore femur measured in lateral view. face (dorsal) view without moving the head FFW Fore femur width: maximum width of between measurements (Fig. 1). Longitudinal fore femur measured in lateral view. alitrunk length measurements (PnL, ML, PpL, FI Femur index: FFW/FFL. PO) were taken in lateral view, parallel to a line HL Head length: maximum length of head (‘measuring axis’) drawn between the anterior- in full face view, from the anterior most point of the pronotal collar and the pos- clypeal margin to the midpoint of a line terior-most point of the propodeal tubercles drawn across the occipital margin (Fig. 3). Alitrunk height measurements (MH, (Fig. 1). PpH) were made perpendicular to the measur- HW Head width: maximum width of head ing axis (Fig. 3). in full face view, excluding eyes in CL Clypeal length: measured in full face workers and queens, including eyes in view (Fig. 1). males (Fig. 1).
FIGS 1-3. Measurements taken during this study (T.pac@ca).1. Full face view. 2. Dorsal view of alitrunk. 3. Lateral view of alitrunk. Revision of Turneriu 103
LES Lateral eye space: distance from lateral RFFL Relative fore"femurlength: FFL/HW. edge of head to lateral edge of eye RES Relative eye spread: ESIHW. measured in full face view (Fig. 1). RLES Relative lateral eye space: LES/HW. MH Mesothoracic height: maximum height RMW Relative mesonotal width: MWIHW. of the mesothorax measured perpen- ROOD Relative occiput-ocular distance: dicular to the measuring axis (see OODIHW. above and Fig. 3). RPO Relative propodeal overhang: PO/HW. ML Mesonotal length: distance from the SI Scape index: SLIHW. pronotal-mesonotal suture to the meta- SL Scape length: length of the first an- notal groove measured parallel to the tennal segment (scape) excluding the measuring axis (see above and Fig. 3). basal radicle (Fig. 1). MW Mesonotal width: minimum width of mesonotum in dorsal view profile (Fig. Genus Turneria Forel 2). OCD Ocular-clypeal distance: distance be- Turneria Forel, 1895: 419 (nec Parent, 1934: tween anterior margin of eye and pos- 127). Type species: Turneriu bidentutu Forel, terior margin of clypeus measured in 1895: 419 (by monotypy). full face view (Fig. 1). 01 Ocular index: EW/EL. Workers of Turneriu may be recognized among OOD Occiput-ocular distance: distance be- the Dolichoderinae by their elongate compound tween posterior margin of eye and pos- eyes (01 0.43-0.66), lack of pilosity on the terior extremity of occipital border dorsum of the head and thorax, dorsal place- measured in full face view (Fig. 1). ment of the propodeal spiracle on the propodeal PI Pronotal width index: PnWIHW. protuberance (Fig. 3) and the configuration of PnL Pronotal length: distance from the an- the declivitous face of the propodeum (concave terior edge of the pronotal collar to the with the propodeal angle rounded and without a pronotal-mesonotal suture measured ridge or pair of spines). Additional non-diagnos- parallel to the measuring axis (see tic characters are as follows. The petiolar scale above and Fig. 3). is present, inclined anteriorly and nodiform. Pn W Pronotal width: maximum width of Colour is either uniform yellowish brown to pronotum, measured in dorsal view black or bicoloured with the head and alitrunk (Fig. 2). yellow and the gaster reddish brown to brown. PO Propodeal overhang: distance from the Erect hairs are limited to the clypeus, the area posterior-most point of the propodeal between the frontal carinae, the mandibles, the tubercles to the posterior-most point maxilla and labium, the coxae and trochanters, of the petiolar insertion measured the fourth gastric tergite and all gastric sternites. parallel to the measuring axis (see The venter of the petiole has suberect hairs. Ap- above and Fig. 3). pressed pubescence covers the majority of the POI Propodeal overhang index: POIPpH . body. The sculpture is weakly (integument PPH Propodeal height: maximum height of shiny) to moderately (integument subopaque) the propodeum measured perpendicular imbricate (Harris, 1979). The anterior clypeal to the measuring axis (see above and margin is entire. The masticatory margin of the Fig. 3). mandible has 5 apical teeth with a denticle be- PPL Propodeal length: distance from the tween the basal-most pair, and 4 or 5 denticles metanotal groove to the posterior- proximally. The basal margin is denticulate and point of the propodeal tubercles the basal tooth is present but weakly developed. measured parallel to the measuring The antennae are 12 segmented. The maxillary axis (see above and Fig. 3). palps are 6 segmented and the labial palps 4 seg- Propedeal length index: PpLIHW. mented. The mid and hind tibia1 spurs are single Propodeal width: maximum width of and pectinate. Total body length is 2.0-3.6 mm. the dorsal propodeal surface in dorsal Queens (based on the four species for which view (Fig. 2). they are known) may be diagnosed by the fol- REL Relative eye length: EL/HW. lowing combination of characters. Petiolar scale 104 Steven 0. Shattuck inclined anteriorly, nodiform and without erect diversity found within Iridomyrmex, a closely hairs dorsally; apical and subapical teeth of related and morphologically similar genus. mandible subequal in size; erect hairs absent on Additional, non-diagnostic queen characters the occiput, propodeum and anterior region of include the following. Head elongate (CI 1.20- the first gastric tergite; third funicular segment 1.30) with three ocelli and the clypeus entire; only slightly longer than broad; and the R-M antennae 12 segmented; maxillary palps 6 seg- crossvein distal of the 2r vein in the forewing mented and labial palps 4 segmented; mandible (sometimes only slightly). Eye and head shape, with 8 teeth, the basal tooth distinct, and the as well as the pilosity of the pro- and meso- basal margin denticulate; wings as in Fig. 21, notum, are of little value because of the extreme with a closed radial cell, two cubital and one dis-
FIGS 4-5. T.arbusta worker (holotype). 4. Full face view. 5. Lateral view of alitrunk. FIGS 6-7. T.hiden/ataworker (lectotype). 6. Full face view. 7. Lateral view of alitrunk. FIGS 8-9. T.co//ina(holotype). 8. Full face view. 9. Lateral view of alitrunk. Revision of Turneria 105
coidal cell; erect hairs present on the mandibles, on the dorsum of petiole and first gastric ceg- labium, all surfaces of the head anterior of the ment. These characteristics occur in Iridom,r- ocelli, the coxae, and the gastric tergites and mex but none are found in the known queens sternites, especially the fourth and fifth pair; of Turneria. The peculiar structure of the pro- erect hairs are present or absent on the pro- podeum of T.butteli (angular with vertical ridges and mesonotum. The mid and hind tibia1 spurs laterally) is presumably what led Forel to assign are single and pectinate. The propodeal spir- this species to Turneria. acles are as in Iridomyrmex (i.e. not displaced T.frenchi Forel (1911: 207, worker from an dorsally as in the Turneria worker). unspecified locality in Australia) is more prob- The male (known for arbusta, bidenfata and lematical. The type material was not found dur- pacifica) is as follows. Head as in Fig. 19; an- ing this study. It is not present in any of the tennae 13 segmented; anterior clypeal margin collections listed above under Depositories, nor entire and angulate; mandibles with a concave in the Institut Royal des Sciences Naturelles de truncation distally and the masticatory margin Belgique (P. Dessart, pers. comm.), which is the smooth, without teeth or denticles; maxillary 'Musee de Bruxelles' mentioned as the type de- palps 6 segmented, labial palps 4 segmented. pository in the original description. No speci- The thorax (Fig. 20) has the scutellurn enlarged mens were encountered during this study which and strongly arched dorsally, the mesothoracic agreed with the original description of this episternum divided medially by a transverse species. However, a single worker specimen suture and enlarged ventrally below the level of identified by W. C. Crawley as T.frenchi was the coxae, and the propodeum broadly rounded found in OXUM. Upon examination, it was but with more or less distinct basal and decliv- found to be an apparently undescribed species of itous faces. The petiole is angular above, with- Stigmacros (Campostigmacros). The specimen out a distinct scale, evenly rounded transversely, Forel described is also apparently a Stigmacros and broadly attached to the gaster. Wings as in rather than a Turneria, but it is not conspecific Fig. 22, with a closed radial cell, one cubital and with the OXUM specimen. Based on the orig- one discoidal cell. Erect hairs are limited to the inal description T.frenchi possesses the follow- clypeus, mandibles, the venter of the petiole, ing characters: eyes convex and placed laterally the gastric sternites and terminal segments, and on head, antennae exceeding occipital border, the volsellae. The entire body is covered with a alitrunk margined laterally, promesonotal fine, appressed pubescence. suture and metanotal groove deeply impressed, declivitous face of propodeum with dentiform Species included in Turneria processes basally, scale notched dorsally, and arbusta spn. integument smooth and shiny. None of these bidentata Forel characters are found in Turneria, but do occur in collina sp.n. specimens of Stigmacros in ANIC, PSWC and dahlii Forel the descriptions of McAreavey (1957). I con- pacifica Forel clude that T.frenchi Forel is a Stigmacros postomma spn. (c0mb.n.). Species removed in this study butteli Forel, to Iridomyrmex c0rnb.n. Key to species of Turneria based on workers frenchi Forel, to Stigmacros c0mb.n. 1 One or two erect hairs present on frontal lobes . 2 The species T.birtteli Forel and T.frenchi - Erect hairs absent from frontal lobes ...... 4 Forel are removed from Tumerio. T.butteli Forel (1013: 419, holotype dealate queen from 2 Body uniformly coloured: yellowish brown to red- dish brown (eastern islands of Papua New Guinea, Tandjong Slamat, Sumatra, Indonesia (ZMHB) Solomon Islands, Vanuatu) ...... dahlii [examined]) is an Iridomyrmex (c0rnb.n.). The - Body strongly bicoloured with head and gaster of following characters support this conclusion. contrasting colour: head yellow. gaster reddish Compound eyes positioned lower (more brown ...... 3 anterior) on head, third funicular segment much 3 Propodeum in dorsal view with the lateral protuber- longer than broad, scale of petiole bluntly ances extending posteriorly (giving a concave ap- pointed dorsally, and long, erect hairs present pearance) so that a line drawn between them would 106 Steim 0. Sharruck
not contact the medial areas; eyes placed relatively - Lateral areas of head with moderate imbricate high on head (ROODc0.26) (Fig. 15); head and sculpturing and with the integument opaque; eye mesonotum short (HLc0.70, MLC0.27) (Papua spread small (ES4.3.S) ...... 5 New Guinea) ...... posromma 5 Propodeum in dorsal view with the lateral protuber- Propodeum in dorsal view without lateral protuber- ances extending posteriorly and with the area be- ances extending posteriorly and with the posterior tween them forming a U-shaped concavity; Clc0.88 margin even or convex across entire width; eyes (Australia) ...... bidentara placed relatively low on head (ROODB0.29) (Fig. - Propodeum in dorsal view either without the lateral 19); head and mesonotum long (HLM.71, protuberances extending posteriorly and with the MLM.29) (Solomon Islands. Vanuatu) . pac(fica posterior margin even or convex across entire width, Lateral areas of head with weak imbricate sculptur- or with the lateral protuberances extending slightly ing and with the integument shiny; eye spread large posteriorly and with the area between them forming (ES>0.35) (Papua New Guinea) ...... collina a very shallow V-shaped concavity; C1>0.86 (Papua New Guinea) ...... arbusra
FIGS 10-12. T.duhlii worker. 10. Full face view (holotype). 11. Lateral view of alitrunk (holotype). 12 Lateral view of alitrunk (Guadalcanal, Solomon Islands). FIGS 13-14. T.pacifica worker (Espiritu Santo, Vanuatu). 13. Full face view. 14. Lateral view of alitrunk. FIGS 15-16. T.postomma worker (holotype). IS. Full face view. 16. Lateral view of alitrunk. Revision of Turneria 107
Turneria arbusta sp.n. (Figs 4, 5) pale yellow. The alitrunk is dark reddish brown, with the legs pale reddish brown and the tarsi Holotype measurements. OOD 0.17, EL 0.26, yellow to reddish yellow. The petiole and gaster OCD 0.13, CL 0.16, HL 0.71, LES 0.06, EW are reddish brown. The ventral region of the 0.14, ES 0.26, HW 0.65, SL 0.47, PnL 0.38, ML petiole is pigmented, not translucent as in 0.29, PpL 0.22, PnW 0.43, MW 0.26, PpW 0.30, bidentata and pacijca. PO 0.03, FFL 0.49, FFW 0.22, MH 0.39, PpH Dkcussion. This species is known from several 0.26, C1 0.91, 01 0.56, REL 0.40, S1 0.73, F1 well-separated collections in Papua New Guinea 0.45, PI 0.67, PpI 0.74, ROOD 0.27, POI 0.12, (Fig. 23). Two series were collected from low- RPO 0.05, RMW 0.40, RLES 0.09, RES 0.40, level vegetation in forested areas. A nest in dead RFFL 0.76. twigs of Premna serratifolia in littoral vegetation Worker measurements (n= 12). OOD near mangrove (8 km N Madang) cantained 0.16-0.22, EL 0.22-0.28, OCD 0.11-0.15, CL males on 29 January. Another nest in dead twigs 0.15-0.19, HL 0.65-0.78, LES 0.03-0.06, EW along the edge of second-growth rain forest and 0.13-0.16, ES 0.23-0.30, HW 0.56-0.68, SL gardens (2 km E Maprik) contained males on 9 0.45-0.52, PnL 0.34-0.41, ML 0.27-0.33, PpL February. A third nest in dead twigs of ?Piper 0.19-0.26, PnW 0.38-0.45, MW 0.22-0.30, sp. along the edge of second-growth rain PpW 0.25-0.33, PO 0.03-0.06, FFL 0.47-0.56, forest (3 km S Wewak) contained males on 15 FFW 0.19-0.23, MH 0.33-0.44, PPH 0.23-0.30, February. CI 0.86-0.93, 01 0.51-0.66, REL 0.38-0.42, Type material. Holotype worker from SI 0.70-0.81, FI 0.39-0.48, PI 0.63-0.69, PpI PAPUA NEW GUINEA: Morobe District, 0.72-0.83, ROOD 0.25-0.32, POI 0.09-0.22, Huon Peninsula, lower Busu River, 5.v.1955, RPO 0.04-0.09, RMW 0.39-0.47, RLES 0.05- lowland rain forest (E. 0. Wilson no. 945) 0.09, RES 0.40-0.46, RFFL 0.73-0.85. (MCZC); 6 paratype workers, same date as Worker diagnosis. REL>0.38, PpI<0.83, holotype (MCZC, BMNH); 11 paratype workers frontal lobes without erect hairs, lateral areas of from PAPUA NEW GUINEA: Northern Dis- head moderately imbricate and with integument trict, 8 km S Kokoda, 800 m (Taylor) (ANIC). opaque, area between propodeal protuberances Other material examined. PAPUA NEW slightly concave to slightly convex. GUINEA: East Sepik District, Ambunti, The clypeus possesses 10-14 erect hairs. The 4"13'S, 142"49'E, 50 m (Ward) (PSWC), Village body colour is uniform reddish brown to dark of Kimbangoa (=Kimbangwa), 2 km E Maprik, brown, with the anterior portions of the head 3"38'S, 143"04'E, 200 m (Ward) (PSWC, and the mandibles slightly lighter, and the SOSC), Cape Moem, 7 km E Wewak, 3"34'S, dorsum of the head slightly darker, especially in 143"41'E, <5 m (Ward)(PSWC, SOSC), 3 km S light-coloured individuals. Wewak, 3'37'5, 143"37'E, 400 m (Ward) Queen description. C1=1.30 (n=l). One or (PSWC, SOSC); Madang District, 8 km N two suberect hairs present on the lateral margin Madang, 5"09'S, 145"48'E, 4 m (Ward) of the head between the posterior edge of the (PSWC, SOSC). eye and the occipital corner; erect hairs absent from the thoracic dorsum. Head, alitrunk, Turneria bidentata Forel (Figs 6,7) petiole and legs (except tarsi) black; tarsi pale Turneria bidentutu Forel, 1895: 419. reddish yellow. Gastric segments 1-3 reddish brown, with the posterior portion of each scler- Worker measurements (n=25). OOD ite pale yellow; segments 4 and 5 dark reddish 0.15-0.21, EL 0.24-0.28, OCD 0.10-0.13, CL brown. 0.13-0.19, HL 0.65-0.75, LES 0.03-0.05, EW Mule description. The following characters are 0.11-0.15, ES 0.24-0.28, HW 0.55-0.66, SL in addition to the generic characterization. The 0.41-0.46, PnL 0.30-0.42, ML 0.27-0.37, PpL antenna1 scape is slightly longer than funicular 0.23-0.30, PnW 0.36-0.43, MW 0.23-0.30, segments 1,2 or 3. The first funicular segment is PpW 0.25-0.29, PO -0.01-0.04, FFL 0.41- enlarged relative to t'le remaining segments, and 0.53, FFW 0.16-0.21, MH 0.31-0.39, PPH nearly spherical. The head is black posteriorly 0.22-0.29, CI 0.80-0.88, 01 0.43-0.57, REL with the clypeus dark reddish brown and the 0.40-0.48, SI 0.70-0.79, FI 0.34-0.48, PI 0.63- mandibles reddish yellow. The antennae are 0.71, PpI 0.87-1.10, ROOD 0.28-0.33, POI 108 Steven 0.Shatturk
-0.04-0.17, RPO -0.02-0.07, RMW 0.42- and rnesonotum. .Head. alitrunk, petiole and 0.47, RLES 0.05-0.09, RES 0.41-0.45, RFFL 'legs black, except its follows: mandibles, scapes, 0.66-0.94. first funicular segments and base of each tibia Worker diagnosis. CIt0.88, KELX.40, dark reddish yellow, and tarsi reddish yellow. frontal lobes without erect hairs, lateral areas of Gastric tergites 1, 2 and 3 reddish black with the head moderately imbricate and with integument exweme anterior, posterior and lateral margins opaque, area between propodeal protuberances yellowish red; sternires 1.2 and 3 yellowish red; concave. segments 4 and 5 black. In lateral profile, the concave region of the Mde description. Characters in addition to the declivitous face of the propodeurn is more generic characterization are as follows. An- rounded than in other species (Fig. 7). The tennal acapc longer than the first or third funicu- clypeus possesses 8-14 erect hairs. Colour varies lar segments, approximately the same length as from uniform dark brown or black to strongly the second. Entire body uniformly reddish bicoloured with the head, alitrunk, legs and yellow, with the head slightly darker posteriorly petiole yellow and the gaster brown. In lighter (near the ocelli), grading to yellow on the mand- coloured individuals, the dorsum of the head ibles. The ventral region of the petiole is without may be slightly darker than the alitrunk. pigment and more or less transparent. Qimn description. Additional characters not Discussion. Body colour in this species is given in the generic description include: CI extremely variable. Populations from the vicin- 1.26-1.28 (n=4). One to about 5 suberect hairs ity of Cairns are uniform dark brown, Mackay- present on the lateral margin of the head be- arca populations are strongly bicoloured. while tween the posterior edge ot the eye and the southern collections from Burleigh He, d.s are occipital corner. Erect- hairs absent from the pro- yellowish brown with the gaster slightly darker.
FIGS 17-18. T.tltrlr/iiqueen (Kitrawat. New Britain. Papua New Guinea). 17. Full face vicw. 18. Lateral vicu of ali t 1-uiik . FIGS l%20. T.ptrc@rr male (Espiritu Santo. Vanuatu). 19. Full f;ice view. 20. Lateral view of alitrunk. Revision of Turneria 109
None of these populations diverge in any of National Park, 23"13'S, 150"38'E, 80 m (Ward) the morphological traits studied, although the (ANIC, PSWC, SOSC), Koala Park, Burleigh Cairns-vicinity specimens average slightly, but Heads, 50 ft (Lowery) (ANIC). insignificantly, smaller. For most traits all collec- tions broadly overlap. The Cairns-vicinity popu- Turneria collina sp.n. (Figs 8,9) lations diverge slightly from the more southern Holotype measurements. OOD 0.21, EL 0.28, ones in two metric traits, MW and PpH, as OCD 0.13, CL 0.20, HL 0.81, LES 0.07, EW follows: MW 0.23-0.27 v. 0.25-0.30 and PpH 0.16, ES 0.35, HW 0.78, SL 0.59, PnL 0.45, ML 0.22-0.25 v. 0 23-0.29. The broad overlapin the 0.35, PpL 0.24, PnW 0.52, MW 0.34, PpW 0.38, ranges of these characters make them of little PO 0.02, FFL 0.62, FFW 0.24, MH 0.54, PpH value in separating these populations. Addi- 0.38, CI 0.96, 01 0.56, REL 0.35, SI 0.75, FI tional collections, especially of queens and 0.38, PI 0.67, PpI 0.64, ROOD 0.26, POX 0.04, males from bicoloured populations, will be RPO 0.02, RMW 0.44, RLES 0.09, RES 0.45, needed to determine the significance of these RFFL 0.80. colour pattern and size polymorphisms. Worker diagnosis. Frontal lobes without erect T.bidentata has been collected from only a hairs, lateral areas of head weakly imbricate and narrow strip along the Queensland coast from with integument shiny, area between propodeal Daintree in the north to Burleigh Heads in the protuberances flat, head and thorax reddish south (Fig. 23). Most sites of known elevation brown and gaster dark reddish brown. The are less than 300 m, although on the Atherton clypeus possesses 22 erect hairs. This species Tablelands it occurs at an elevation of about differs from all other known species in the fol- 700 m. It occurs on parkland and rain forest trees lowing measurements: ES, PpW, MH and PpH. (standing or recently felled), including Tristania It also differs from any given species by an addi- conferta. Nests have been found in dead twigs of tional three to sixteen metric characters. Lantana sp., Macaranga sp. and Terminalia Discussion. This species, known from a single catappa. Queens and males have been collected specimen, is one of the most distinct in the from a nest on Green Island, Queensland, on 20 genus. It is the largest species and differs from January. A twig nest in Lantana sp. occupied other Turneria by ten to twenty-three metric about 1 m of non-contiguous sections near the characters. (Part of this apparent uniqueness twig nest of a Camponotus (Colobopsis) species. may be due to the small sample size. Additional The nest, collected 16 January, 6 km SSE of material may show overlap in some characters Atherton, contained about 429 workers, 2 alate with known species.) The collection site, at an and 1 dealate queen, and numerous eggs, larvae elevation of 1600 m, is unusually high for the and pupae. genus. The next highest collection site for Type material. Two syntype workers from Turneria is 1000 m and most collections are AUSTRALIA: Queensland, Mackay (M. G. below 100 m. Turner) (MHNG) [examined]. The lower spe- Type material. A single worker from PAPUA cimen on the pin is here designated as NEW GUINEA: Western Highlands District, LECTOTY PE . Nondugl, 1600m,9.vii.1955 (Gressitt) (MCZC). Other material examined. AUSTRALIA: Queensland, Cooper Creek, near Daintree Turneria dahlii Forel (Figs 10-12) (Feehan) (ANIC), 12 km ENE Daintree, 16"13'S, 145"25'E, 10 m (Ward) (ANIC, Turneria dahlii Forel, 1901: 17; Emery, 1912: 21 PSWC), Cairns (Taylor)(ANIC), Cairns district unjustified emendation to T.duhli); Wilson, (Lae) (MCZC), Green Island, 16"46'S, 1962: 17 (queen described). 145"58'E (Ward) (ANIC, PSWC, SOSC), Rus- Worker measurements (n=47). OOD sell River at Bellender Ker Landing, 5 m 0.18-0.26, EL 0.20-0.27, OCD 0.10-0.14, CL (ANIC), Wongabel State Forest, 6 km SSE 0.14-0.23, HL 0.67-0.86, LES 0.04-0.08, EW Atherton, 17"19'S, 145"30'E, 720 m (Shattuck, 0.11-0.16, ES 0.27-0.34, HW 0.61-0.80, SL Ward) (ANIC, PSWC, SOSC), 1 mile E East 0.48-0.63, PnL 0.31-0.45, ML 0.24-0.37, PpL Palmerston School, c. 600 ft, 17"37'S, 145"50'E 0.20-0.30, PnW 0.40-0.57, MW 0.24-0.34, (Taylor)(ANIC), Giru (Lowery)(ANIC), Mac- PpW 0.23-0.32, PO 0.04-0.08, FFL 0.49-0.65, kay (R. E. Turner) (BMNH), Mt Jim Crow FFW 0.19-0.26, MH 0.36-0.51, PpH 0.25-0.37, 110 Steven 0.Shaiiuck
CI 0.88-0.98, 010.53-0.63, REL 0.29-0.37, SI 1.28), the relative scape length (SI 0.83), and the 0.72-0.85, FI 0.32-0.47, PI 0.61-0.80, PpI uniform dark brown colour are more similar to 0.71-1.07, ROOD 0.26-0.35, POI 0.10-0.31, known duhlii queens (CI 1.25, 1.26; SI 0.80, RPO 0.04-0.12, RMW 0.37-0.46, RLES 0.05- 0.84; uniform colour) then those of pacifica (CI 0.13, RES 0.41-0.50, RFFL0.73-0.86. 1.20-1.21; SI 0.76-0.79; bicoloured (n=4)). Worker diagnosis. EW
FIG. 21. Fore and hindwings of T.pacifica queen (Espiritu Santo, Vanuatu). FIG. 22. Fore and hindwings of T.pacificn male (Espiritu Santo. Vanuatu). Revision of Turneria 111 strongly rounded (Fig. 12). The basal face of the (Gressitt) (MCZC), Bainings, St Paul's, 350 m propodeum (also in profile) varies from moder- (Gressitt) (MCZC); West New Britain District, ately arched (Fig. 11) to more or less flat (Fig. Kimbe district, Kavui subdivision (Brown) 12). Both of these characters are fairly labile and (MCZC); New Britain, Yalom, loo0 m (Noona extremes can be found within single nest series. Dan. Exp. 61-62) (ANIC). SOLOMON Collection sites occur between 0 and 1000 m, ISLANDS: Rennell, Hutuna (Bradley) with most below 100 m. Specimens are found on (BMNH); Bellona (Greenslade) (ANIC); trees in rain forests,.with one collection from Malaita, Dala (Greenslade) (ANIC); Guadal- Hydnophytum. Wilson (1962) found this canal: Tenam Ridge (P. J. M. Greenslade) species, together with T.pacifica, to be one of (ANIC), Mt Austin (Greenslade) (ANIC). the dominant arboreal ants on Espiritu Santo. VANUATU: Espiritu Santo, 8 km SW Lugan- Type material. Holotype (unique syn- ville (Wilson) (MCZC); Malekula, Ounua type) worker from PAPUA NEW GUINEA: (Cheesman) (BMNH) . Bismarck Archipelago, Kabakaul (MHNG) Turneria pacifica Mann (Figs 13, 14, 17-22) [examined]. Other material examined. PAPUA NEW Turneriapacifica Mann, 1919: 361; Wilson, 1962: GUINEA: Manus District, Manus Island, near 17 (queen described); Wheeler, 1934: 178 Lundret Village, 2"03'S, 147"11'E, 100 m (distribution: Matema Bay, Santa Cruz (Huxley) (ANIC); East New Britain District, Island, Solomon Islands). Rabaul (Krauss) (BPBM), Keravat, 60 m Worker measurements (n=15). OOD
0
%l 6,1 b b 10
0
20
V orbusta H bidentoto 'I 0 collina I dahlii A pocifica + postomma I 1 30 > I50 I60 170 FIG. 23. Distribution of Turneriu studied. 1 I2 Stcww 0. Shatturk
0.20-0.23, EL0.22-0.26, OCD 0.11-0.14, CL Discussion. This species occurs in the Solo- 0.15-0.22, HL 0.71-0.83, LES 0.06-0.08, EW mon Islands and Vanuatu (Fig. 23) and is com- 0.11-0.14, ES 0.29-0.34, HW 0.66-0.76, SL pletely sympatric with dahlii. Wilson (1962) 0.50-0.59, PnL 0.34-0.43, ML 0.27-0.33, PpL found it, along with dahlii, to be one of the 0.22-0.27, PnW 0.41-0.50, MW 0.27-0.34, dominant arboreal ants in lowland rain forests PpW 0.27-0.30, PO 0.04-0.08, FFL 0.53-0.61, on Espiritu Santo. FFW 0.19-0.23, MH 0.37-0.49, PpH 0.27-0.34, Type material. Holotype worker from SOLO- CI 0.YO-O.Y5,010.51-0.59, REL0.32-0.35, SI MON ISLANDS: Santa Cruz, Grasciosa Bay 0.75-0.86, FI 0.32-0.40, PI 0.59-0.68, PpI (Mann) (USNM) [examined]. 0.78-0.91, ROOD 0.29-0.33, POI 0. IS-0.27, Other material examined. SOLOMON IS- RPO 0.07-0.11, RMW 0.39-0.44, RLES 0.08- LANDS: San Cristobal, Warahito River 0.11, RES 0.42-0.46, RFFL 0.77-0.83. (Greenslade) (ANIC); Santa Cruz Islands, Worker diagnosis. REL<0.35, frontal lobes Matema Island, Mohawk Bay (Willow) with 1 or 2 erect hairs, lateral areas of head (CASC); Santa Cruz, Grasciosa Bay weakly imbricate and with integument shiny, (=Graciosa Bay) [type plus a second specimen, area between propodeal protuberances flat, see above] (Mann) (USNM). VANUATU: Es- bicoloured with yellow head and thorax and piritu Santo, 8 km SW Luganville (Wilson) dark brown to black gaster. The clypeus pos- (MCZC) . sesses between 10 and 16 erect hairs. As discus- sed above, pacifica and dahlii are very similar Turnerid postomma sp.n. (Figs 15, 16) and the only significant difference found in this study was the contrasting yellow and black col- Holotype measurements. OOD 0.14, EL 0.24, our of pacifca, while dahlii is uniform yellow- OCD 0.10, CL 0.16, HL 0.64, LES 0.04, EW brown to black. 0.13,ES0.27,HW0.62,SL0.49,PnL0.40,ML In the USNM is a specimen bearing an ident- 0.25, PpL 0.23, PnW 0.42, MW 0.26, PpW 0.34, ical locality label as the type specimen (also in PO 0.05,FFL 0.51, FFW 0.19, MH 0.38, PpH the USNM). This specimen is conspecific with 0.31, CI 0.97, 01 0.56, REL 0.39, SI 0.79, FI the type and was apparently collected at the 0.38,PI0.68,PpI0.69,ROOD0.23,POI0.15, same time. Mann (1919) clearly states he RPO 0.08, RMW 0.41, RLES 0.06, RES 0.43, described this species from a single specimen, RFFL 0.82. therefore this second specimen should not be Worker measurements (n= 15). OOD considered part of the type series. 0.13-0.17, EL 0.22-0.24, OCD 0.10-0.14, CL Queen description (Figs 17, 18, 21). Addi- 0.13-0.16, HL 0.60-0.70, LES 0.03-0.06, EW tional characters not given in the generic de- 0.12-0.14, ES 0.24-0.29, HW 0.58-0.67, SL scription include: numerous erect hairs present 0.45-0.51, PnL 0.33-0.41, ML 0.23-0.27, PpL on the lateral margin of the head between the 0.20-0.24, PnW 0.38-0.45, MW 0.24-0.27, posterior cdge of the eyc and the occipital cor- PpW 0.30-0.37, PO 0.04-0.09, FFL 0.46-0.54, ner, and on the pro- and mesonotum. Body FFW 0.16-0.21, MH 0.37-0.43, PpH 0.28-0.33, bicoloured with the head, alitrunk, legsand peti- CI 0.92-0.97, 010.55-0.63, REL 0.35-0.39, SI ole yellow to reddish yellow and the gaster dark 0.75-0.83, FI 0.33-0.41, PI 0.64-0.70, PpI brown to black. For additional characters and 0.63-0.71, ROOD 0.21-0.26, POI 0.12-0.27, discussion sce duhlii under Queen description. RPO 0.06-0.13, RMW 0.41-0.45, RLES 0.04- Mule description (Figs 19,20,22). Characters 0.09, RES 0.42-0.43, RFFL 0.79-0.88. not listed in the generic characterization are as Worker diagnosis. REL 0.35-0.42, OOD< follows. Antenna1 scapc longer than the first or 0.16, frontal lobes with a single erect hair, lat- third funicular segments, approximately the eral areas of head weakly imbricate and with same length as the second. Head reddish brown, integument shiny, area between propodeal pro- darker near the ocelli; mandibles, antennae and tuberances concave, bicoloured with yellow legs ycllow; clypeus and frontal area reddish yel- head and thorax and dark brown gaster. The low; thorax and dorsum of petiole reddish yel- clypeus possesses between 14 and 20 erect hairs. low with the gaster slightly darkcr; ventral area This species superficially resembles pacifica or of petiole lacking pigment and more or less bidentata in colour pattern but can easily be diag- transparent. nosed with the above characters. Revision of Turneria 113
Discussion. This species is known from two taxa overlapped less than 5% they were treated collections from lowland rain forests in Papua as separate states. Where ranges overlapped by New Guinea (Fig. 23). Its biology and nesting more than 5%, the character was scored as the habits are unknown. same, or if more than one state of another taxon Type material. Holotype worker from was incorporated, as polymorphic. Multiple PAPUA NEW GUINEA: Morobe District, Di- gaps were found in three characters (HL, CI, diman Creek, Lae, 8.v. 1955, lowland rain forest PpI) which allowed discrimination of different (E. 0. Wilson no. 979) (MCZC); 8 paratype subsets of taxa. These three characters were workers, same data as holotype (MCZC, each divided into two new characters (HL1, ANIC). HL2: CI1, CI2; PpIl , PpI2) and coded independ- Other material examined. PAPUA NEW ently. The resulting character states are given GUINEA: Morobe District, lower Busu River, in the appendix and the states for each taxon in Huon Peninsula (Wilson) (USNM, MCZC, Table 1. Range analysis resulted in two classes of BMNH). character states: discrete, with gaps between ad- jacent taxa; and polymorphic, with ranges over- lapping several taxa. Discrete states were coded Character analysis and species relationships for analysis using standard binary coding (0, 1). Polymorphic characters, however, possess three Species of Turneria are morphologically very states (two alternate discrete states and poly- similar to one another and possess a limited morphic) and thus could not be coded in the number of discrete ,characters. Because of this, normal manner. Three alternate models of char- the following analysis relies primarily on con- acter state evolution were investigated and the tinuous metric characters. The relationships advantages and disadvantages of these are dis- among species of Turneria were determined by a cussed below. cladistic analysis of nine quantitative (see below) In the first model of character state evolution and four qualitative characters. See the Appen- (Model 1) characters were coded as 0,l and un- dix and ‘Measurements and indices’ for an ex- known (polymorphic). Minimum length (most planation of the characters and character states. parsimonious) trees were then found using the The quantitative characters were coded for Alltrees option of PAUP ver. 2.4.1, developed analysis using two alternate procedures: division by D. L. Swofford. This option performs an based on character ranges (‘range analysis’) and exhaustive search of all possible trees and allows character means (‘generalized gap-coding’). The multiple derivations and reversals of character resulting species-level phylogenies for both of states (Wagner assumptions). In Model 2, these procedures are discussed below. polymorphic characters were recoded into The first procedure, range analysis, attempts multi-state characters with states 0, 1 (polymor- to divide the nine metric characters into states phic) and 2. As in Model 1, PAUP was then used which minimize overlap between taxa. States to find most-parsimonious trees. Model 3 used were established by examining the range of val- the following pattern of state evolution: origina- ues for each taxon. If the character ranges of two tion of polymorphism (state 0 to l or state P),
TABLE 1. Dataset derived from range analysis.
HHPCCRPPRPRRPHFC LLnIIEppOOPLrdro 12L1 2 LI I01 OEoSn 1 12D Stct arbusta Plll POlOPOOPllll bidenrata PllllooooPPlollP collina 0000011101101011 dahlii PPlPOlPOOOOPlOOl pacijica OPllolPOooOPlooo poslomma 1llP011ll0010000 Froggattella OP0O0111O??00000 I14 Steven 0.Shattuck followed by the retention of polymorphism and The generalized gap-coded data was analysed the subsequent fixation of a single state (state P using PAUP as discussed above. The dataset was to state 0 or 1) where necessary. This model was analysed twice, once with all characters hav- proposed by Felsenstein (1983) and is im- ing equal weights, and once with characters plemented in his Dolpenny program of Phylip weighted inversely proportional to their number ver. 3.1. The results of these models are dis- of states. Inverse weighting insures all char- cussed below. acters of equal importance in tree generating To determine the effect of polymorphic data algorithms (Goldman, 1988). The weights used on generated trees, additional analyses were are listed in Table 2. The results of this proce- performed using PAUP with datasets containing dure are discussed below. only characters with (ten characters) and with- In all analyses the resulting trees were rooted out (six characters) polymorphic taxa. These re- using an outgroup (Froggattella) and characters sults were then used to evaluate the various were not polarized. (Thus in Tables 1 and 2, ‘1’ character coding procedures outlined above. or ‘2’ may represent the ancestral condition and The second procedure for character coding, ‘0’ the derived condition.) Froggattella appears generalized gap-coding, uses the character mean to be closely related to Turneria based on the for each taxon and a ‘discriminant criterion’ (the position of the propodeal spiracles and anterior weighted average standard deviation of each inclination of the petiolar scale in the worker, taxon) to construct subsets of taxa which are not and the nodiform petiolar scale and apical man- identifiably different. These subsets are then as- dibular teeth in the queen. Additionally, Eisner signed character states suitable for phylogenetic (1957) found the proventiculi of Turneria, Frog- analysis. A given taxon may be placed in its own gattella, as well as Iridomyrmex, to be identical. subset (if it differs from all other taxa), a subset While Turneria and Froggattella are closely re- with one or more taxa (if all have similar means) lated, the nature of their relationship to or in more than one subset (if it is similar to two Iridomyrmex has yet to be determined, and other taxa which differ from each other by more therefore Froggattella was chosen as an out- than the discriminant criterion). The subsets are group to root trees. then converted into states (M-codes of Goldman, 1988) based on the pattern of subsets across taxa. The taxon means and states, and Results character discriminant criteria for the sixteen characters analysed with this procedure are The results of the three range analysis coding given in Table 2. A detailed treatment of the schemes are as follows. Coding polymorphic generalized gap-coding procedure is presented states as unknown (Model 1) and utilizing the by Goldman (1988). full dataset resulted in a single, fully resolved
TABLE 2. Dataset derived from generalized gap-coding. Values are M-codes (see text), followed by original character means in parentheses. Abbreviations are as follows: arb=arbusta, bid=bidentata, col-collina, dah=dahlii, pac=pacifca, pos=postomma, Frog= Froggattella, d=discriminant criterion, w =weight.
arb hid col dah Pac POS Frog d W
HL 3(0.72) 2(0.70) 6(0.81) 4(0.74) 4(0.75) O(0.64) 6(0.85) 0.039 0.167 PnL 2(0.37) O(0.31) 4(0.45) 2(0.38) 2(0.39) 2(0.37) 6(0.54) 0.054 0.167 CI 2(0.89) O(0.84) 6(0.96) 4(0.93) 4(0.92) 6(0.95) 8( 1.05) 0.020 0.125 REL S(0.40) lO(0.44) 4(0.35) 2(0.33) 2(0.34) 6(0.38) O(0.24) 0.015 0.100 PPI 4(0.77) 8(0.99) 2(0.64) q0.80) 6(0.85) 2(0.64) O( 0.40) 0.061 0.12s ROOD 3(0.28) 4(0.30) 2( 0.26) 5(0.31) 5(0.31) O(0.24) 7(0.44) 0.018 1.143 POI 2(0.15) O(0.07) O( 0.04) 4(0.20) 4(0.20) 3(0.16) Unknown 0.048 0.250 RPO 2(0.06) O(0.03) 0(0,02) 4( 0.09) 4(0.08) 3( 0.08) Unknown 0.020 0.250 RLES 2( 0.08) l(0.07) 4( 0.09) 4(0.10) 4(0.10) O(O.06) 6(0.20) 0.012 0.167 01 3( 0.57) O(0.51) 3(0.56) 3(0.58) 2(0.55) 4(0.58) 2(0.55) 0.032 0.250 SI 2(0.75) 210.74) 3(0.75) 5(0.79) 4(0.78) 5(0.79) O(0.71) 0.024 0.200 RES O(0.43) l(0.44) 2(0.45) 2(0.44) 2(0.44) O(0.42) 4(0.61) 0.014 0.250 Revision of Turneria 115 With generalized gap-coding, data was analysed both weighted (inverse weighting, see above) and unweighted. Analysis with weighting resulted in a single tree with the taxa in the fol- lowing order (from the base of the tree): post- omma, pacifica, dahlii, collina, arbusta and bidentata. This is similar to the trees found with Models 1 and 2, differing only in the placement of collina as a sister group to arbusta+bidentata rather than basal within the entire genus. Analysis without weighting resulted in two trees, W with a consensus tree as follows: collina and FIG. 24. Phylogenetic relationships of Turneria pacifica+dahlii in an unresolved trichotomy species (see text for explanation). (Frog.=Froggat- with postomma and arbusta+ bidentata. This tella. ) tree shares the placements of collina and ar- busta+bidentata in common with Models 1 and 2, but places postomma as a sister group to ar- tree similar to Fig. 24, but with pacifica placed busta+bidentata, rather than as a sister group to basal to dahlii. Analysis with the ten polymor- pacifica +dahlii+arbusta +bidentata. phic characters resulted in seventy equally par- simonious trees, and gave a consensus tree which did not resolve any taxa, thus providing no Discussion information on species relationships. Model 2, which allows multiple derivations and losses of In this study, nine analyses were performed: polymorphism, resulted in two equally par- seven utilizing three models of range analysis simonious trees when utilizing the full dataset. A with three datasets, and two utilizing general- consensus tree for these is shown in Fig. 24. ized gap-coding with two weighting schemes. Analysis using the ten polymorphic characters These analyses resulted in three fully resolved, resulted in eleven equally parsimonious trees, five partially resolved and one unresolved tree. with a consensus tree which placed collina basal Each method has strengths and weaknesses to the remaining Turneria species, but gave no which need to be considered when evaluating other information. The six non-polymorphic the resulting trees for that model. A discussion characters gave nine equally parsimonious trees, of each model follows. with a consensus tree which was unresolved ex- Range analysis model 1 (coding polymorphic cept for the placement of arbusta and bidentata states as unknown) may allow tree generating al- as sister taxa. (Since this dataset contains no gorithms unjustifiable flexibility in character polymorphic characters, these results are inde- placement and result in trees containing little pendent of the model of character state evolution or no information. This was the case when us- used.) Model 3, which requires a single origina- ing the dataset containing only characters with tion of polymorphism followed by subsequent polymorphic taxa, which resulted in seventy loss, resulted in a single most-parsimonious tree equally parsimonious trees and a completely un- when utilizing the full dataset. It differed from resolved consensus tree. However, adding six the trees found with Models 1 and 2 in having ar- non-polymorphic characters resulted in a single busta+ bidentata located basal of postomma, fully resolved tree, suggesting that this method rather than as a sister group to dahlii orpacifica. may be appropriate if the ratio of polymorphic to When analysed with the polymorphic characters non-polymorphic characters is not too high. alone, two equally parsimonious trees resulted. Range analysis model 2 (allowing multiple de- The consensus tree for these has collina basal to rivations and losses of polymorphism) utilizes the remainder of the species, and bidentata in an the widely accepted Wagner assumptions for unresolved trichotomy with dahlii+pacijica and character state evolution. When utilizing the full postomma+arbusta. This is the only analysis dataset, it resulted in two trees differing only in which did not place arbusta and bidentata as the placement of one taxon. When using only sister taxa. polymorphic characters, this method showed 116 Steven 0.Shuttuck only slightly better resolving abilities than analysis and generating most-parsimonious trees Model 1, with one taxon being placed basally to utilizing Wagner assumptions about character the remaining unresolved species of the genus. state evolution. Polymorphic states can be The similarity of these results with those of treated as unknown if they constitute a small Model 1 (with the full dataset) combined with proportion of the total data matrix, but should the slightly better resolution with the poly- be treated as part of a linear transition series morphic dataset suggest this model may be ap- (with two alternate discrete states and propriate if the number of qualitative characters polymorphism) if the number of non-polymor- is limited. phic characters is limited. Using these methods, Range analysis model 3 (requiring a single Fig. 24 best summarizes the species relationships origination of polymorphism followed by sub- within Turneria. sequent loss) was developed specifically to analyse polymorphic characters. With the full Conclusion dataset, the single resulting tree was similar to the results of Models 1 and 2, but differed in the Polymorphic characters are a potentially import- placement of two taxa. When utilizing the poly- ant source of phylogenetic data, although they morphic characters alone, two trees resulted present special problems for a cladistic analysis. which differed significantly from trees found Removal of these characters may result in lost with other methods, including this method with information, and can seriously compromise the the full dataset. For example, it pfacedpostomma estimated phylogeny. One method for coding and urbustu as sister taxa, a relationship not polymorphic characters, generalized gap-cod- found with any other method. In addition, the ing, utilizes the character mean and a single rather dramatic change in tree topology when weighted mean variance value. Unfortunately, utilizing the two datasets suggests problems with polymorphic taxa will have larger ranges for the underlying assumptions of this model. It may these characters compared to non-polymorphic be unjustifiable to limit the polymorphic con- taxa and thus larger variances. The potential loss dition to a single origination and may be more of this variance information may adversely affect appropriate to allow multiple derivations as with the resulting phylogeny. Models 1 and 2. Another method of treating polymorphic data Generalized gap-coding with inverse weight- is to code the polymorphic condition as either ing resulted in a single tree differing from the re- unknown, or as intermediate in value between sults of Models 1 and 2 in the placement of one two alternate monomorphic states (i.e. part of a taxon. Without weighting characters, two trees linear transformation series). Treating polymor- resulted which differ significantly from Models phic states as unknown may give acceptable re- 1 and 2. This suggests that the use of weights sults if the number of polymorphic characters is is appropriate, especially considering that gen- low compared to non-polymorphic characters in eralized gap-coding can result in characters the dataset. If few non-polymorphic characters with numerous states (e.g. ten states for REL, are available, this method may reduce available see Table 2). An additional potential problem information and result in unjustifiable flexibility with generalized gap-coding is the loss of of character placement by tree-generating al- polymorphic information caused by a single vari- gorithms. In this situation, polymorphic charac- ance value being placed around the mean of ters should be treated as part of a transformation each taxon. Use of a single variance value is ap- series and allowed multiple derivations and propriate if all taxa have similar variances, but is losses. While this seems the most appropriate less valid for unequal variances. Taxa with method it also treats polymorphic and discrete polymorphic characters will have larger ranges, states as equally important. It seems likely dis- and consequently larger variances. This infor- crete states possess more reliable phylogenetic mation is lost in the discriminant criterion used information than polymorphic states and con- for taxon subset construction, which is an aver- sequently should be allowed more weight during age variance estimate across all taxa. tree generation. Finally, allowing a single ori- An evaluation of all models tested leads to the gination of polymorphism followed by sub- conclusion that polymorphic characters are best sequent loss seems unjustifiably restrained in not analysed by coding them into states using range allowing multiple originations of the polymor- Revision of Turneria 117
phic condition as permitted by the previous Harris, R.A. (1979) A glossary of surface sculpturing. model. California Department of Food and Agriculture. Laboratory Services, Entomology. Occasional Papers, 28, 1-31. Mann, W.M. (1919) The ants of the British Solomon Acknowledgments Islands. Bulletin of the Museum of Comparative Zoology, 63,275-391. I wish to thank the following for loans of McAreavey, J.J. (1957) Revision of the Genus Stig- macros Forel. Memoirs of the National Museum of material: A. Arakaki (BPBM), C. Besuchet Victoria, 21, 7-64. (MHNG), B. Bolton (BMNH), F. Koch Parent, 0. (1934) Additions a la faune ethiopienne (ZMHB), M. W. Moffett and S. P. Cover (Diptkres: Dolichopodides). Bulletin. Sociktk En- (MCZC), C. O'Toole (OXUM), W. J. Pulawski tomologique d'Egypte, 18,112-138. Wheeler, G.C. & Wheeler, J. (1974) Supplementary (CASC), D. R. Smith (USNM) and R. W. studies on ant larvae: Simopone and Turneria. Taylor (ANIC). The comments of an anony- Journal of the New York Entomological Society, mous reviewer are greatly appreciated. 1 also 82, 103-105. thank P. Dessart for his assistance in searching Wheeler, W.M. (1934) Formicidae of the Templeton for type material. P. S. Ward and M. K. Smith Crocker Expedition, 1933. Proceedings of the California Academy of Sciences, (4),21,173-181. provided valuable encouragement and com- Wilson, E.O. (1959) Some ecological characteristics ments throughout this project. of ants in New Guinea rain forests. Ecology, 40, 437-447. Wilson, E.O. (1962) The ants of Rennell and BeIlona Islands. Natural History of Rennell Island, British References Solomon Islands, Vol. 4, pp. 13-23.
Eisner, T. (1957) A comparative morphological study of the proventriculus of ants (Hymenoptera: For- Accepted 9 October 1989 micidae). Bulletin of the Museum of Comparative Zoology, 116,437-490. Emery, C. (1912) Fam. Formicidae. Subfam. Dolichoderinae. In: Wytsman, P. Genera Insec- forum Fax. 137,l-50. Appendix Felsenstein, J. (1983) Methods for inferring phylogenies: a statistical view. In: J. Felsenstein Explanation of cladistic character states (ed.). Numerical Taxonomy. NATO AS1 Ser., Character states: HL1>0.71 (0), <0.70 (1); Vol. G1, pp. 315-334. Forel, A. (1895) Nouvelles fourmis d'Australie, recol- HL2>0.81 (0), ~0.78(1); PnL>0.45 (0), <0.45 ties i~ The Ridge, Mackay, Queensland, par M. (1); CI1>0.96 (0), <0.96 (1); CI2>0.88 (0), Gilbert Turner. Annales de la Sociktk Entomo- <0.88 (1); REL>0.38 (0), <0.39 (1); PpIl> logique Belgique, 39,417-428. 0.87 (0), <0.83 (1); PpI2>0.72 (0), <0.64 (1); Forel. A. (1901) Formiciden aus dern Bismarck- Archipel, auf grundlage des von Prof. Dr. F. Dahl ROOD>0.26 (0), ~0.26(1); POI>O.O9 (0), gesammelten materials. Mitteilungen aus der <0.04 (1); RP0>0.04 (0), <0.02 (1); RLES> Zoologischen Museum in Berlin, 2, 4-37. 0.09 (01, <0.09 (1); Prot' concave (0), convex or Forel, A. (1911) Ameisen aus Java beobachtet und flat (1); HdScZweakly (0), moderately (1) imbri- gesammelt von Herrn Edward Jacobson. 11. Theil. cate; Frnt3 present (0), absent (1); C014 Notes from the Leyden Museum, 33, 193-218. Forel, A. (1913) Ameisen aus Sumatra, Java, Malacca bicoloured (0) , uniform (1). und Ceylon. Gesammelt von Herrn Prof. v. Buttel-Recpen in den Jahren 1911-1912. 'Area between lateral corners of the propodeal Zoo logkche Jahrbucher. Systematik, Okologie und tutlercles. Geographie der Tiere, 36, 1-148. -Relative development of sculpture on lateral areas Goldman, N. (1988) Methods for discrete coding of of head. morphological characters for numerical analysis. 'Erect hairs on frontal lobes. Cladistics. 4, 59-71. 4Body colour .