and P. Santschi Systematic status of Plectroctena mandibularis Smith conjugate (: Formicidae: )

1 1 2 Villet L. McKitterick & H.G. Robertson M.H. ,

F. Because there has Plectroctena maiuiilnilans Smith is the type species of Plectroctena Smith. to as- been some doubt about its distinctness from P. conjitgnta, several techniques were used several series contained sess the systematic status of the two species. Most crucially, colony or the workers of both phenotypes, and where these series included queens males, not related to those of the distinguishing feature of these specimens was consistently workers. Queens, males and workers did not manifest qualitative differences between the between the two. The taxa, and morphological variation was continuous putative morpho- for workers of the taxa arose from allometric logical basis (funicular index) distinguishing in males was related to but no variation. Putatively diagnostic colour variation latitude, could be correlated with distribu- simple pattern of morphological variation geographical of P. niandilnilaris. tion. Plcctivctt'iia conjugate! is therefore considered a junior synonym southern Key words: , systemati rphometrics, biogeography,

INTRODUCTION

whether is a distinct cannot be The Plcctroctcnn F. Smith, 1858, was de- cunjitgata species a decision scribed to accommodate P. nmndilntlnrif Smith, settled satisfactorily at present, and must await the of more of all 1858. This species has been difficult to distinguish amassing specimens the of three castes'. In three males of from P. conjitgntn Santschi, 1914, workers particular, only were available to Bolton for examination. which were distinguished from those of P. inaiuii- conjugata variation and small have 1'iilarif by being smaller and having funicular Geographic samples confounded assessment segments 3-5 shorter than wide, rather than consequently systematic of P. as the behaviour of these southern African longer than wide (Table 1). Queens conjugate! and, taxa is attention apparently differ from P. mandibnlaris in being attracting contemporary (Villet et al. Peeters & Crewe Villet 1989, smaller, with proportionally shorter mandibles; 1984; 1988; 1991), clarification of their status is essential. males have a black gaster, while P. mandibularif, systematic males are yellow, orange-brown or reddish MATERIAL AND METHODS (Table 1). Material in the Grahamstown Arnold (1926) synonymized P. conjugata with P Albany Museum, National iimndibulnris on the basis that An examination of a (AMGS), Durban Museum (DMNH), author Collection of Pretoria Transvaal specimen [of conjugntn] sent to me by the , (SANC), Pretoria and South African [Santschi] shows that it does not differ from the Museum, (TMSA), is in the Town was examined. small variety of mandibularif, which found Museum, Cape (SAMC), from the same nest a nest coastal regions of the eastern part of the Cape Specimens (i.e. series) assumed to to the same Sev- Province. The species becomes progressively were belong species. and a eral nest series contained different castes and larger the further it is from the coast, Caste was determined the of complete transition series can be found from the sexes. by presence ocellar in the form of two at the sites smallest (Durban, etc.) to the largest or typical vestiges pits nmndilnilaris, such as is found in Rhodesia'. Bolton of the lateral ocelli in queens (Bolton 1974). (1974) treated the taxa as distinct, with the cniviil Additional collections mentioned in the list of syn- T am not convinced onyms are the Natural History Museum, London (BMNH), Museum National d'Histoire Naturellc, >3.0 mm Worker head length <3.0 mm 0.38-0.52 mm Worker eye length 0.32-0.38 mm >4.0 mm Queen head length <4.0 mm Queen mandibular index <83 >83 >2.9 mm Queen scape length <2.8 mm wide As or than wide Funicular segments 3-5 Annular, shorter than long longer anterior half of Covering whole pnia Striae on gula Covering gu Black Red-brown to oranuc Male gaster

of intranidal variation was Paris (NHM), Museum of Comparative Zoology, examination important in this Based on the traits Cambridge, Massachusetts (MCZ), Musee Royal study. distinguishing nest series and listed in Table 1, 13 (28 %) of the 47 de 1'Afrique Centrale, Tervuren (MRAC), of both P. inandibulnris and /'. Naturhistorisches Museum, Basel (NMB). contained workers

1 Instances of this can be seen in , Quantitative variation in males, workers and conjugatn. Fig. where the of was decided on queens was determined phenetically using identity specimens 18 the basis of the mean funicular index of the eolom morphometric measurements of 180 workers, variables series. Several worker from P. conjugate queens and 40 males. The morphometric specimens and defined colonies have funicular indices above 100, and were primarily those suggested by even more from P. innndihiihin^ Bolton (1974). In the case of workers and queens specimens have indices below 100. the the measurements were head length (HL), man- colonies Revising critical FI did not resolve the contradiction. Similar dibular length (ML), scape length (SL), maximum dorsal unresolvable contradictions arose from using ocular diameter (OD), petiole length (PL), of funicular head or eye length to identify (presumably i petiole width (DPW), length segment length 4 in related workers. 4 (FSL) and width of funicular segment (FSW); mandi- FI head as a measure males they were head length, head width, Regressing against length allometric of size 1) a significant ble length, eye length, mandible width, clypeus body (Fig. yielded 2 = = 394.6; P = 0.000; R 0.69). The width and mesonotum length. The pigmentation relationship (F scored as use of ratios such as funicular index to taxa of gasters of males was orange/yellow separate ratio ex- or black Bolton is based on the assumption that a (1), brown (2), brown-tipped (3) (4). implicit several of size. That this is not (1974) also proposed and defined ratios, pressed shape independent true is evident in this case. the most important of which are the mandibulo- always (Bookstein 1989) = for Four of the six associated with workers cephalic index (MI ML/HL), separating queens index = had traits (mandibulo-cephalic queens, and the funicular (FI FSL/FSW), morphological head and/or length (Table 1 )) for discriminating between workers (Table 1). indices, length scape them in a different from Measurements were log-transformed to that would place species five of the six males with linearize allometric variation, and analysed using the workers. Similarly, and black P. coujngnta) were associ- principal component analysis regression. gasters (putative variation and ated with P. mandibularis females. Colony member- Geographic patterns in quantitative corre- therefore contradicted most putative pigmentation were examined by canonical ship and as traits. lation analysis, using latitude longitude distinguishing measures of geographical location. Morphometric variation: workers RESULTS AND DISCUSSION Since other discriminatory traits could exist, it was also necessary to further test for the subdivi- characters that Colony membership sion of the sample of using can be have useful for other members of the As specimens from the same nest series proved eenus assumed to belong to the same species, an (Bolton 1974). Head length (mm)

Fig. 1. Allometric regression of cular index on hea<

P. conjugala (

Table 2. Pearson's correlations of the first two components of a principal comr. with each of the morphometric variables and with the funicular index for work

Logiofhead length) Logio(mandible length)

Log ic (scape length) Log, (eye length) Logio(petiole length) lorsal petiole widt Logi (funicular se llogio(funicular segment Funicular index

Principal component analysis carried out on the index was significantly correlated with both com- measurements of 180 workers showed only one ponents, although the degree of correlation with = = significant axis (Ei 7.3; E: 0.2), which summa- the second component was poor, and its correla- rized 91.3 ';('. of the variation. The second axis ex- tion with the first component was weaker than " plained only a further 2.85 of the variance and that of the original variables. tailed to separate the taxa (Fig. 2). The separation In morphometric studies, the first axis of a prin- of the traditional taxa along the first axis was quite cipal component analysis of log-transformed data marked, although discrete groups were not is generally an expression of size, and the second formed (Fig. 2). The component weights of the first axis can be interpreted as a size-free expression of axis (Table 2) were all similar, and correlated highly shape if all the original variables show similar, high and variable Ta- significantly with every original ( correlations with the component scores li le 2). Only eye length and funicular segment (Bookstein 1989). Since they do so in this study length were significantly, but weaklv, correlated (Table 2), the second axi^ suggests that workers of with the second component liable 2). Funicular the two taxa are the same shape. Component 1

The geographical distribution of samples of The first axis differentiated specimens on size, but workers covered most of the eastern half of characterized only 68.4 % of the total variation. southern Africa an of axis (Fig. 3), allowing assessment The second had a high weighting (0.8) for eye Arnold's (1926) hypothesis of geographical size length, but failed to distinguish the taxa. The variation. The first axes of a canonical correlation remaining details of the analysis were similar to analysis showed strong positive loadings for both latitude and longitude, but a high score for petiole length only (Table 3); the loadings of the remain- ing measurements were of similar, small magni- of tude and mixed sign (i.e. variously positive and negative). The morphological axis showed a sepa- ration of the taxa (Fig. 4) similar to that on the first axis of the principal component analysis (Fig. 2), Eigenvalue 0,3827 with P. mcmdibnlaris generally scoring higher than Canonical con 0.6186 P. the canonical correlation conjiigatn. However, Wilk's lambda 0.4955 0.8026 was and no marked trend only 0.62, geographical Chi-square 121.12 37.92 emerged (Fig. 4). The second axes showed weaker correlation (Table 3), with a strong loading for latitude but not longitude, and emphasis on funic- Coefficients ular and length. This segment length eye pattern Log 10 (head length) -0.36162 0.67113 the results of the parallels principal component Log, (mandible length) 0.24693 -0.98377 analysis, that the second com- indicating principal Log 10 (scape length) -0.34650 -0.09114 ponent summarizes variation, and geographical Log IO (eye length) -0.38172 that that variation is small relative to the implying Log, (petiole length) 2.34857 overall variation in the sample. Log ic (dorsal petiole width) -0.60227

Log 10 (funicular segment length; -0.65837 variation: Morphometric queens Log, (funicular segment width) 0.50363 -0.13773 The 18 formed a diffuse cluster queens single, Latitude 1.20386 0.96842 the first axes when two of the principal compo- 1.54448 -0.04097 nent analysis were plotted against one another. Villet et a/.: Systematic status of Plectroctena species

P. worKers 01 Plectroctena mandibulahs ( + ) and conjugata (x). Fig. 3. Geographical distributions ot samples

FI and canonical correlation of their those concerning the analysis of workers. analysis morphology correlated with and coloration with location was MI were significantly but weakly geographical = The first axes the first and second axes respectively (n\ 0.65; significant (Table 4). emphasized with small r = indices of latitude and coloration, very M! -0.51), implying that they are poor gaster of mixed for the other variables general morphology. For these reasons, no loadings sign so that a latitudinal trend in distinguishing features were identified. (Table 4), only strong from most of colour was revealed. Such trends are not Samples of queens were collected gaster correlation rare within of southern African and southeastern Africa (Fig. 5). Canonical species wasps of axes bees ct d. 1998; C.D. Eardley 1983, pers. analysis found no significant sets relating (Steele = = F.W. Gess, comm.), which suggests a morphology to distribution (Ai 0.13; A: 0.47; comm.; pers. effect. P > 0.10 for both sets of axes). common underlying environmental

Morphometric variation: males TAXONOMIC CONCLUSION of a Intranidal variation contradicts A plot of the first two axes principal compo- morphological the traditional means of P. nmndi- nent analysis of the morphological measurements distinguishing from P. since of of 40 males showed only one spheroidal cluster, Inthms conjugatci, representatives occur in the same nest. This conclu- with the two taxa thoroughly mixed. The first axis both taxa may sion is the lack of distinc- had an eigenvalue of 2.9, accounting for 36.2 % of supported by categorical tion between them in the the variation. It was not uniformly or strongly principal component of males. in the workers and correlated with the original variables. Abdominal analysis Only queens the was there a of differentiation between the pigmentation and clypeus width dominated degree is attributable to size and its second axis with absolute weightings of 0.51, but taxa, and this body on allometric variation. There was also no did not correlate strongly with the associated effect of the taxa component scores. coherent geographical differentiation besides latitudinal variation in the gaster colour of males. No categorical variation in the structure of Geographical variate

correlation of the and Fig. 4. Scatter diagram of the first canonical axes derived from canonical analyses morphology geographical location of samples of workers of Plectroctena mandibularis (_i) and P. conjugate ().

mandibularis Smith the male genitalia could be found (Robertson, Plectroctena 1914 is Plectroctena mandibularis F. 1858: 101, unpubl.). Plectroctena conjugntn Santschi, Smith, pi. 1-5. and consequently relegated to synonymy of P. mandi- 7, figs. Syntype queen (ergatoid) bularis F. Smith, 1858. The synonymy of P. mandi- male, SOUTH AFRICA: Natal, Durban, bularis is therefore as follows. Gueinzius (BMNH). to Ponern caffrn Spinola, 1851: 53 (attributed 4. canonical correlation of Table Results of analysis Klug). (Nomcn nudiim; material referred to morphology and sample location for males of Plectroc- mandibularis by Roger 1861: 41). tena mandibularis and P. conjugata. 74. Plectroctena caffrn st. major Forel, 1894: Holotype queen (not worker as stated), MOZAMBIQUE: Delagoa, P. Berthoud (NHM). (Synonymy by Emery 1899: 469.) Eigenvalue Plectroctena minor st. conjugata Santschi, 1914: 8. Canonical correlation Syntype workers and queen (ergatoid), Wilk's lambda SOUTH AFRICA: Natal, Stamford Hill, Chi-square Charlestown, 30. iv. 1905, and Zululand, I. Tragardh (MCZ, MRAC, NMB). (Raised to species by Santschi 1924: 166; maintained as Coefficients species by Bolton 1974: 326). Syn. n. Abdomen colour 0.93981 0.19334 Plectroctena mandibularis var. Integra Santschi, -0.30739 0.20642 Log 10 (head length) 1924: 161. Syntype worker, KENYA: Nairobi, Log width) 0.17712 -1.15098 10 (head Wa Kikongo et Masai, 1904, Ch. Alluaud; Log (mandible length) 0.07094 0.07119 10 and syntype male, KENYA: Bura, Wa Taita, length) -0.13843 -0.18159 Log, (eye 1904, "Ch. Alluaud (NMB). (Synonymy by Log (mandible width) 0.15749 0.01367 10 Bolton 1974: 330.) Log (clypeus width) -0.18025 0.20776 10 Plectroctena mandibularis st. strigosa var. Log (mesonotum length) -0.20999 -0.03976 10 strialiveiitns Stitz, in Santschi 1924: 162. (Un- 1.03211 0.46879 Latitude available name; material referred to Longitude 0.07253 1.13126 mandibularis bv Bolton 1974: 330).

tht ACKNOWLEDGEMENTS j. Keulder (Rhodes University) prepared referee T. Crouch (DMNH), C.D. Eardley (SANC), F.W. figures. An anonymous generously pro- Gess (AMGS) and R. Toms (TMSA) kindly ar- videci details of the synonymy. The study was the Foundation foi ranged the loan of material in their care; A. Cook financially supported by helped with laboratory work, and S. Abraham and Research Development and Rhodes University.

REFERENCES

ARNOLD, G. 1926. A monograph of the Formicidae of du Zululand recoltes par le Dr. I. Tragardh; avec un Ivar South Africa; appendix. Annals of the Soutli African appendice; Notes biologiques par Tragardh. Museum 23: 191-295. Gotel'orgsKungliga Vetetukapsoch VitterHets-Sniwinlles 15: 117. BOLTON, B. 1974. A revision of the ponerine genus Handlingar Sjatte Foljden F. 1924. du Plectroctena. Rerue Plectroctena F. Smith (Hymenoptera: Formicidae). SANTSCHI, Revue genre '.'- i T" Bulletin of the British Museum (Natural History), Sitisse dc

F. 18...,. .. , 30: _ ..., , Entomology 311-338. SMITH, , v Part 6. formicidae. Lon- BOOKSTEIN, F.L. 1989. 'Size and shape': a comment on Collection of the British Museum, semantics. Systematic Zoology 38: 173-180. don. EARDLEY, C.D. 1983. A taxonomic revision of the genus SPINOLA, M. 1851. Compte rendu des Hymenopteres du de Xylocopa Latreille (Hymenoptera: Anthophoridae) in inedits provenants voyage entomologique M southern Africa. Entomology Memoir of the Department Ghiliani. \\emone tie In Accatli'inie de la Science* dc Torrino 19-94. of Agriculture of the Republic of South Africa 58: 1-67. (2)13: H.R. & EMERY, C. 1899. Fourmis d'Afrique. Annales de la Societe STEELE, G.R., VTLLET, M.H., HEPBURN', 1998. Clinal variation Entomologiquc de Bclgique 60: 459-504. RADLOFF, S.E. morphometric FOREL, A. 1894. Abessinische und andere afrikansche in wild honeybees (Hymenoptera: Apidae) in South Alfred Africa. and Distribution 4: 17-25. Ameisen, gesammelt von Herrn Ingenier Ilg, Diversity von Herrn Dr. Liengme, von Herrn Pfarrer Missionar VILLET, M.H. 1989. A syndrome leading to ergatoid P Berthoud, Herrn Dr. Arth. Muller etc. Mittelunge queens in ponerine ants (Hymenoptera: Formicidae). 23: 825-832. Sclmvize Entomologiesche Gesellschaft 9: 64-100. Journal of Natural History PEETERS, C.P. & CREVVE, R.M. 1988. Worker laying in VILLET, M.H. 1991. Colony foundation in Plectroctena F. and the evolution of the absence of an ergatoid queen in the ponerine ant mandibularis Smith, ergatoid genus Plectroctena. South African Journal of Zoology 23: queens in Plectroctena (Hymenoptera: Formicidae). 78-80. lounial of \ntural History 25: 979-983. Ameisen. Berliner C.P. & CREVVE, R.M. 1984. The ROGER, J. 1861. Die Pom-ra-artigen VILLET, M.H., PEETERS, of a in four of Entomologische Zeitsclirift 5: 1-54. occurrence pygidial gland genera SANTSCHI,' F. 1914. Meddenlanden fran Goteborgs ponerine ants (Hymenoptera: Formicidae). Journal of 19: 413-416. Musei Zoologiska Afdeling no. 3. Fourmis du Natal et the Georyia Entomological Societu

Accepted 6 Novel