Out of the Neotropics: newly discovered relictual species sheds light on the biogeographical history of spider ants (Leptomyrmex, Dolichoderinae, Formicidae) Brendone Boudinot Rodolfos Probst Carlos Roberto F. Brandão Rodrigo Feitosas Philip S. Ward Systematic Entomology 2016 41 658 671 [821,1035,462,483] Insecta Formicidae Leptomyrmex Animalia Hymenoptera 5 663 Arthropoda species relictus sp. nov.    Typematerial.   Holotype worker. BRAZIL, Goiás: Niquelândia, Serra da Mesa, 14∘20′39.5′′S, 48∘10′21.1′′W,  16.xi.2013, cerrado, inside nest of  Cornitermes silvestrii(  D. E. Oliveira, R. G. Santos & T. Carrijo) (sample code DEO2013-030; specimen code CASENT0106501) [ MZSP].    Paratypeworkers(w) and males(m). Same data and colony as holotype; sample code R06[ 1 w MZSP, 1 w( CASENT0730400)  1 m UCDC]; sample code TC08[ 1 w DZUP, 1 w MZSP]; sample code DEO2013-030[ 1 w LACM,  1 m MZSP]. Additional material examined (nontype): 12 w, 6 m, 3 larvae, and 2 pupaefrom 6 localities in Goiás and Tocantins, Brazil( Fig. 2).   Diagnosis.With diagnostic characters of genus from Shattuck (1992), notably with medial hypostomal notch (synapomorphy of genus) and lacking pterostigmal appendage in male (synapomorphy of macro-  Leptomyrmex). Distinguished from all  Leptomyrmexby the following characters: (i) short, thick, erect setae present on the head capsule, mesosoma (legs included) and metasoma ( Figs 1A, 3A, B); (ii) clypeal setae comparatively more numerous ( Figs 1E, 3B); (iii) anterior clypeal margin strongly convex ( Figs 1E, 3B); (iv) lateral hypostoma weakly flanged; (v) worker with convexity subtending subapical mandibular tooth ( Fig. 1E); (vi) worker petiole lateromedially narrow; (vii) worker gaster strongly lateromedially compressed; (viii) male scape elongate (SI 1.26–1.40 versus 0.19–1.05 inother  Leptomyrmex); (ix) male wing with five closed cells (costal, basal, subbasal, submarginal 1, marginal 1) ( Fig. 5E); (x) male petiole dorsal margin concave ( Fig. 3A); (xi) male petiolar spiracles situated on tubercles ( Fig. 3A); (xii) several features of male genitalia ( Fig. 3C–K; see Appendix S2 for description). Characters 1–5, 9–11 and those of 12 are unique in the genus.   Biology.Although males of  L. relictushave been collected as far back as 1956 – remaining unidentified for over half a century – workers were only recently discovered in 2013 from nest samples of  Cornitermes silvestriiEmersonand in 2012 from epigaeic and arboreal pitfall traps in typical cerrado (Brazilian savannah) habitat. It is unknown whether workers are nocturnal or diurnal, but hypothetical nocturnal habits and/or their  Camponotus-like habitus may account for the low encounter rate of this species. It is interesting to note that at least three of the specimens were captured in arboreal pitfall traps, suggesting that workers may forage both on the soil and in the vegetation strata. The collection of workers ( Fig. 1A, E) and brood (Figure S1) inside termitaria suggests that  L. relictusmay live in association with  Cornitermes silvestrii.One termite nest measuring 1.30 mwide and about 40–50 cmtall inhabited by  L. relictuswas found in northeast Goiás state under a tree, and was also occupied by another ant species,  Camponotus blandus(Smith, F.), which is known to be a facultative termite inquiline ( Gallego-Ropero & Feitosa, 2014). The other termite nest with  L. relictusspecimens was found in southeast Tocantins state, along a trail edge. Unfortunately, no data are available on how the  Leptomyrmexoccupy the termitaria or if they were sharing galleries with the termites. A potential association with  C. silvestriiis compelling as the termitaria of  C. silvestriiare large, environmentally-stable clay mounds and may house several other ant and termite species (see Appendix S2). In addition to describing and comparing the sociometric and sociogenic patterns ( Tschinkel, 1991, 2011) of  L. relictuswith the Australian species, future studies should test the association of the species with  C. silvestrii.   Phylogeny.Phylogenetic analyses of the 10-gene dolichoderine data matrix yielded results congruent with those of Ward et al.(2010)( Fig. 4). In particular, we consistently recovered all four dolichoderine tribes, with Tapinominias sister to all other Dolichoderinae.  Leptomyrmex relictusis positioned as sister to the six Australian  Leptomyrmexwith strong support: Bayesian posterior probability (PP) of 0.96, and ML bootstrap support (MLBS) of 93%. There is maximum support (PP 1.00, MLBS 100%) for the Australian  Leptomyrmexand for the micro- and macro- clades within it. Our results differ from those of Lucky (2011)for the internal topology of  Leptomyrmexin that, with our sampling,  L. unicoloris sister to  L. erythrocephalus, rather than  L. geniculatus+  L. rufithorax. A sister-group relationship between  Leptomyrmexand the (  ForeliusEmery+  DorymyrmexMayr) clade is also sustained (PP 1.00, MLBS 74%). Finally this larger clade of (  Leptomyrmex+ (  Forelius+  Dorymyrmex)) is sister to all other Leptomyrmecini, as in Ward et al. (2010). Notably,  L. relictusis a long-branched taxon; because its divergence from the most recent common ancestor of the crown  Leptomyrmex, it is apparent that  L. relictushas undergone considerable morphological (especially genitalic) and molecular evolution.   Fig. 5.Results of BEAST and LAGRANGE analyses, and comparative morphology of dolichoderine wing venation. A, Chronogram of the Dolichoderinae,  Leptomyrmex relictus  sp.n.bolded; B,  L. burwelli: complete (unreduced) wing venation; C,  Chronoxenus javanus: loss of crossvein 2rs-m with retention of other veins; D,  Dorymyrmex insanus(male): loss of Rsf2-3 in addition to loss of other abscissae; E,  L.relictus  sp.n.: loss of Mf2-3; F,  L.fragilis: loss of Mf2-3, 2r-rs (note: pterostigmal appendage not illustrated); G, †  L.neotropicus: loss of Rs + M, Mf2-3, 2rs-m, Mf4+, 2r-rs; H, †  Leptomyrmula maravignae: loss of Mf2+, 2rs-m, 2r-rs; I, †  Usomyrma mirabilis: loss of Rs + M with retention of 1m-cu and loss of Mf2-3. In part A blue bars are 95% confidence intervals; coloured circles near nodes are inferred inherited ranges, with the top half corresponding to the upper branch and bottom half with the lower branch; coloured circles next to terminals represent biogeographical coding; asterisks indicate calibrated nodes; bracketed letters to the right of terminal names indicate wing venation reduction pattern occurrence, with letters corresponding to the wing patterns B–F. Note that 1m-cu presence is variable for most taxa displaying loss pattern E, and some  Linepithemaspecies have lost Mf2-3. Figures H and I redrawn from Emery (1891) and Dlussky et al.(2014), respectively. Palaeo., Palaeocene; Oligo., Oligocene; Pl., Pliocene; Q, Quaternary.  Divergence dating.Two alternate analyses were run treating †  Chronomyrmexas a stem dolichoderine, reflecting uncertainty of stem or crown placement of †  L. neotropicuswith respect to extant  Leptomyrmex. Dating results obtained were robust to alternate placements of †  L. neotropicus. The chronogram from the stem  Leptomyrmextreatment is presented in Fig. 5. Mean ages between the two treatments for the clades listed in Table S3 differed by an average of 0.21 Ma, and 95% highest probability densities (hpd) differed by an average of 0.26 Ma at the older and 0.53 Ma at the younger ends, respectively. These findings are in accord with results of Ward et al.(2010)and Lucky (2011)(Table S3), although less so with the latter which incorporated fewer outgroups and loci. Only the dolichoderomorph crown age differed notably from Ward et al.(2010), being 11 Ma younger. This difference is attributable to the modified dolichoderomorph root node calibration used herein which is a diffuse normal distribution, not restricting the minimum age of the node to 100 Ma with the unplaceable yet putative aneuretine fossil †  Burmomyrma.   Biogeographical history.The LAGRANGE analysis was run with a single outgroup taxon (  Aneuretus simoniEmery) ( Fig. 5), and all results are presented in Table S4. A Neotropical origin was recovered with high support (relative probabilities >0.80) for the Leptomyrmecini, the  Leptomyrmex+ (  Dorymyrmex+  Forelius) clade and  Leptomyrmex( Fig. 5A). The crown ages of  Leptomyrmexand the Australasian clade are estimated to be 46 Ma (95% hpd 53–37 Ma) and 28 Ma (95% hpd 40 to 18 Ma), respectively (Table S3, calibration 2). We therefore infer dispersal to Australia between those two time periods. This range expansion probably happened before the breakup of the Antarctic land bridge, estimated to have occurred about 35 Ma for Australia and Antarctica, and about 30 to 28 Ma for Antarctica and South America ( Lawver et al., 1992; Sanmartín & Ronquist, 2004). Another notable result is an inferred Neotropical origin for the Dolichoderini(comprising solely  DolichoderusLund) ( Fig. 5A). Only four nodes had relative probabilities <0.33 for inferred ancestral ranges: the crown and root Dolichoderinae, the BDL ( Bothriomyrmecini+ Dolichoderini+ Leptomyrmecini) clade, and the Bothriomyrmecini. All other relative probabilities were >0.66, except the  Liometopum+  AptinomaFisher+  TapinomaFoersterclade, which had a relative probability between 0.33 and 0.66. Further sampling within the Bothriomyrmeciniis expected to increase the relative probabilities of inferred ancestral ranges for the tribe as well as the BDL clade.   Results and evaluation of fossil record. Dlussky et al.(2014)claimed that †  Usomyrma, †  Leptomyrmulaand  Leptomyrmexform a natural group based on the putatively synapomorphic ‘reduction of the central vein sections and cells.’ This is refuted by the retention of complete wing venation in the species of micro-  Leptomyrmex, which is nested well within the genus. Moreover, the venation reduction patterns of †  Usomyrma,  Dorymyrmex+  Foreliusand  Leptomyrmexare not homologous. †  Usomyrmahas lost Rs + M and Mf2-3; the former is autapomorphic and the latter has occurred at least five times in the Dolichoderinae( Fig. 5). At least two reductions have occurred in  Leptomyrmex( Fig. 5) as the micro- clade retains complete wing venation, whereas  L. relictus, the macro-  Leptomyrmexand †  L. neotropicushave differing reduction patterns.  Leptomyrmex relictushas lost 1 m-cu and Mf2-3, †  L. neotropicushas lost Rsf2+, Mf2 + m, 2rs-m, 1m-cu and 2r-rs, whereas macro-  Leptomyrmexhave lost Mf2-3, 2r-rs (variable) and 1m-cu, and retained Rs + M, Rsf2+ and Mf4+. Other characters claimed by Dlussky et al.(2014)to support a relationship between †  Usomyrmaand †  Leptomyrmexare, by their explicit statement, either plesiomorphic or autapomorphic; these are thus uninformative of relationships by definition. Furthermore, based on examination of a paratypemale, †  Usomyrmalacks a newly discovered genitalic synapomorphy of the Leptomyrmecini: a dorsal sclerotic bridge between the penisvalvar bases. The sclerotic bridge is anteroposteriorly broad and fused with the basimeres in  L. relictus( Fig. 2C, F), and is narrowed anteroposteriorly in the Australasian clade. The strap- or thread-like condition of the bridge in the Australasian clade is a synapomorphy, whereas fusion of the bridge to the basimeres in  L. relictusis autapomorphic. Regarding †  Leptomyrmulaspecifically, the only known specimen may be lost (G.M. Bargossi, University of Bologna, Italy, personal communication), but it has fortunately been illustrated ( Emery, 1891) (reproduced in Fig. 5H). The pattern of vein abscissa loss is most similar between †  Leptomyrmulaand the macro- clade ( Fig. 5F), with the venation being clearly different from the micro-  Leptomyrmex( Fig. 5A) and  L. relictus( Fig. 5E), which have complete or relatively complete wing venation, respectively. Homology of the loss pattern between †  Leptomyrmulaand the macro-  Leptomyrmexis unlikely, however, given the distinctions of the venational patterns. Forewing characters of †  Leptomyrmuladiffer as follows from the macro-  Leptomyrmexclade: costal cell broad and short (vs. absent, with secondary sclerotization); free R lacking along the anterior wing margin (vs. continuing almost to wing apex); pterostigmal appendage absent (vs. present); 2rs-m + Mf4+ absent (vs. usually present); basal cell narrow and Mf1 subparallel to Sc + R + Rs (vs. broad, Mf1 and Sc + R + Rs clearly converging); free Cu short and diverging from M + Cu in the basal third of the wing (vs. about midlength); and anal vein elongated after crossvein cu-a (vs. anal vein strongly reduced or absent). Moreover, the somatic morphology of †  Leptomyrmulais considerably divergent, including the following differences: head capsule dorsoventrally compressed; compound eyes small and anteriorly-situated; mandibles small; flagellomeres short; pronotum dorsoventrally tall; mesonotum flat; and limbs short (metatibia not exceeding metasomal apex). Neither †  Leptomyrmulanor †  Usomyrmahave the key macro-  Leptomyrmexclade forewing synapomorphies of the pterostigmal appendage ( Lucky & Ward, 2010) and the absent costal vein with secondary sclerotization. We therefore consider †  Leptomyrmulaand †  Usomyrma incertae sedisin the Dolichoderinae. The placement of †  Leptomyrmulain the Dolichoderinaeitself could be questioned because the genitalia, as illustrated in Emery (1891), do not conclusively demonstrate a synapomorphic condition of dolichoderine genitalia: enlarged basimere and reduced telomere ( Yoshimura & Fisher, 2011; Boudinot, 2015). Study of worker and male †  L. neotropicusspecimens revealed presence of the hypostomal notch, a synapomorphy of the genus almost completely unique in the Dolichoderinae, except for the enigmatic leptomyrmecine  AnillidrisSantschi(B.E. Boudinot, personal observation), confirming placement of the fossil within the genus. Presence of the notch could not be evaluated in either †  Leptomyrmulaor †  Usomyrma; the ventral head surface of the paratype†  Usomyrmaexamined is obscured by a fracture within the amber. It is plausible that †  L. neotropicusis more closely related to  L. relictusthan to the Australasian clade based on genitalic characters, although the placement of the Dominican amber fossil should be tested in a ‘total-evidence’ framework ( Ronquist et al., 2012b). 1275090574 inside nest of Cornitermes silvestrii 2013-11-16 MZSP D. E. Oliveira & R. G. Santos & T. Carrijo Brazil -14.344306 Niquelandia -48.1725278 Goias 5 663 DEO2013-030, CASENT0106501 1 1 holotype 1275090573 MZSP, UCDC, DZUP, LACM 5 663 code R06, CASENT0730400, code TC08, DEO2013-030 7 2 paratype