Improved Phylogenetic Analyses Corroborate a Plausible Position of Martialis Heureka in the Ant Tree of Life

Improved Phylogenetic Analyses Corroborate a Plausible Position of Martialis heureka in the Ant Tree of Life

Patrick Ku¨ ck1*, Francisco Hita Garcia2, Bernhard Misof1, Karen Meusemann1 1 Zoologisches Forschungsmuseum Alexander Koenig (ZFMK), Bonn, Germany, 2 Department of Entomology, California Academy of Sciences, San Francisco, California, United States of America

Abstract Martialinae are pale, eyeless and probably hypogaeic predatory ants. Morphological character sets suggest a close relationship to the ant subfamily Leptanillinae. Recent analyses based on molecular sequence data suggest that Martialinae are the sister group to all extant ants. However, by comparing molecular studies and different reconstruction methods, the position of Martialinae remains ambiguous. While this sister group relationship was well supported by Bayesian partitioned analyses, Maximum Likelihood approaches could not unequivocally resolve the position of Martialinae. By re-analysing a previous published molecular data set, we show that the Maximum Likelihood approach is highly appropriate to resolve deep ant relationships, especially between Leptanillinae, Martialinae and the remaining ant subfamilies. Based on improved alignments, alignment masking, and tree reconstructions with a sufficient number of bootstrap replicates, our results strongly reject a placement of Martialinae at the first split within the ant tree of life. Instead, we suggest that Leptanillinae are a sister group to all other extant ant subfamilies, whereas Martialinae branch off as a second lineage. This assumption is backed by approximately unbiased (AU) tests, additional Bayesian analyses and split networks. Our results demonstrate clear effects of improved alignment approaches, alignment masking and data partitioning. We hope that our study illustrates the importance of thorough, comprehensible phylogenetic analyses using the example of ant relationships.

Citation: Ku¨ck P, Hita Garcia F, Misof B, Meusemann K (2011) Improved Phylogenetic Analyses Corroborate a Plausible Position of Martialis heureka in the Ant Tree of Life. PLoS ONE 6(6): e21031. doi:10.1371/journal.pone.0021031 Editor: M. Thomas P. Gilbert, Natural History Museum of Denmark, Denmark Received January 19, 2011; Accepted May 17, 2011; Published June 24, 2011 Copyright: ß 2011 Ku¨ck et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The work of BM has been supported by the DFG (www.dfg.de) grant MI649/6-2, and PK is supported by a doctoral fellowship financed by the DFG, SPP 1174 (WA530/33), which is gratefully acknowledged. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected]

Introduction [1,3,4], but contradict previous morphological studies, which assumed that ancestral ants were larger, more wasp-like, epigaeic Recently, a spectacular and rare new subfamily of ants was foragers with well-developed eyes [6–9]. Therefore, the phyloge- described from the Brazilian Amazon with new implications for netic position of Martialinae and Leptanillinae within the ant tree the ant tree of life. The monotypic subfamily, Martialinae was of life still awaits a clear resolution. characterized by a single worker that shows remarkable morpho- Rabeling et al. (2008) [1] presented a Bayesian tree with logical features [1]. It is a small, blind, pale, and most likely resolved single inter- and intra subfamily relationships and hypogaeic predator that lives either in the leaf-litter stratum or proposed Martialinae as the earliest branch (posterior probability directly within the soil. Some morphological characters, such as 0.91) within the ant tree of life. Recent studies have shown that the absence of eyes and frontal lobes, fully exposed antennal Bayesian analyses tend to overestimate the potential signal within sockets, and a flexible promesonotal suture, indicate a closer data and provide high support values, even if the data is relationship to the also small, eyeless, subterranean, and predatory completely uninformative [10,11]. Furthermore, Bayesian ap- ant subfamily, Leptanillinae [2]. Other characters, like a strongly proaches show a much higher type I error rate (the possibility reduced clypeus and long forceps-like mandibles, justify the that erroneous conclusions will be drawn more often), especially establishment of a taxon Martialinae [1]. More important, this in the case of model misspecification [11]. Bayesian posterior new subfamily was presented as a putative sister group to all other probability values are substantially higher than corresponding extant ants on the basis of the molecular analyses of three nuclear bootstrap values [10–13]. Suzuki, Glazko & Nei [10] showed in genes, the small and large nuclear subunits 18S and 28S rRNA simulation studies that Bayesian support values ‘‘can be and elongation factor EF1aF2 [1]. Previous molecular studies had excessively liberal when concatenated gene sequences are used’’. proposed the subfamily Leptanillinae as a sister group of all other Bootstrap values are in general more conservative and more extant ants [3–5]. The proposed sister group relationship of reliable in assessing the robustness of phylogenetic trees which leptanillines suggested in these studies, as well as the one presented should be preferable in phylogenetic analyses [10,11,13]. for Martialinae by Rabeling et al. (2008) [1], is of high significance Therefore, we suggest that topologies inferred with Maximum for a better understanding of ant relationships and ground plan Likelihood (ML) analyses in combination with a sufficient characters. These results strongly support the scenario of a small, number of bootstrap replicates provide a more realistic picture eyeless, and hypogaeic predator as an ancestor of modern ants of the underlying signal.

PLoS ONE | www.plosone.org 1 June 2011 | Volume 6 | Issue 6 | e21031 Improved Analyses for the Ant Tree of Life

We re-analysed the data of Rabeling et al. (2008) [1] using topologies of our masked data sets significantly outperformed H0. partitioned and unpartitioned ML approaches with a sufficient Both AU tests of the masked and the masked-partitioned data set number of bootstrap replicates. Despite the mentioned criticisms on significantly rejected H0 (masked: p,0.0001; masked-partitioned: Bayesian analyses, we additionally conducted comparable Bayesian p = 0.046). Leptanillinae as the first split within the ant tree of life analyses to see whether any of our Bayesian topologies support the was also supported by our split network analyses. Both split relationships found by Rabeling et al. (2008) [1], especially with networks (masked and unmasked) showed less conflict for respect to deep splits. For alignment masking we applied the Leptanillinae as the first split than for Martialinae (see Figure S1). software ALISCORE. Recent studies have shown that alignment Relationships of poneroids and formicoids. None of our masking of positions that can not be aligned unambiguously is topologies recovered a clade poneroids, except the Bayesian strongly recommended to improve the signal-to-noise ratio in topology derived from the unmasked data set (0.86 bpp, see multiple sequence alignments prior to tree reconstruction. Several Figure S4). Further, all ML and Bayesian topologies failed to resolve automated software tools have been developed [14–18] that offer a the relationships between Agroecomyrmecinae, Amblyoponinae, more comprehensible alignment masking than a manual exclusion Paraponerinae, and Proceratiinae. Conflicting signal among these of sites. ALISCORE is a parametric masking approach that subfamilies is seen in both split networks, but the masked network identifies randomised alignment sections by using a Monte Carlo shows less conflict (Figure S2b). In contrast to our unmasked data, resampling within a sliding window [17,18]. The approach assumes all masked approaches resolved a (Ponerinae, formicoids) clade with that the score of inaccurate and ambiguous alignment sections will weak bootstrap and high Bayesian support values (masked- not be distinguishable from randomly similar aligned sequences. unpartitioned: 57% bs, 0.97 bpp; masked-partitioned: 68% bs, Therefore, ALISCORE compares the score of originally aligned 1 bpp; Figures 2 and 3, Table 1, and Figures S6 and S7). A sequences with scores of randomly drawn sequences of similar formicoid clade was maximally supported in all topologies (100% character composition. ALISCORE has been successfully tested bs, 1 bpp). both in simulations [17] and on real data sets [18], and has been Within formicoids, a dorylomorph clade was recovered in all our used in recent molecular phylogenetic studies [19–23]. trees (100% bs, 1 bpp; Figures 1, 2, 3, Table 1 and Figures S2, S3, S4, S5, S6, S7). Four of six topologies suggested a clade Results dorylomorphs+formicoids. However, in the ML masked-unpartitioned topology, the placement of dorylomorphs remained Alignment masking, number of bootstrap replicates and unresolved. In the unmasked Bayesian topology, a clade dorylo- likelihood scores morphs+Pseudomyrmecinae was present, but with weak support Alignment masking remarkably improved data structure, which (see Figure S4). Concerning the relationships between dolichoder- is visualised by comparing split networks derived from the omorphs, Myrmeciinae, and Pseudomyrmecinae, we did not obtain unmasked and masked alignments. The split (NeighborNet) an unequivocal resolution from any topology. The relationships network [24–26] from the masked alignment obviously showed between Formicinae, Myrmicinae and ectaheteromorphs were not less conflict than the split network from the unmasked alignment, resolved by our ML topology of the unmasked data set, whereas the especially within subfamilies of formicoids. Nevertheless, conflict- trees of both masked approaches showed weak node support for a ing signal is obvious, e.g. within poneroids or dorylomorphs (see clade Myrmicinae+ectaheteromorphs (unpartitioned: 73% bs; Figure S1). partitioned: 67% bs). This clade was also resolved in all Bayesian We determined the number of sufficient bootstrap replicates for topologies with moderate support (see Figures S5, S6, S7). our ML analyses using the ‘bootstopping criterion’ according to Pattengale et al. (2010) [27] (see method section). Our unmasked Discussion data set converged after 2,400 bootstrap replicates, our masked- unpartitioned data set after 3,400 bootstrap replicates, and the A clade Leptanillinae + all remaining ant subfamilies is highly masked-partitioned data set after 4,100 bootstrap replicates supported in all our ML and Bayesian analyses. This result is applying the Weighted Robinson-Foulds (WRF) distance criterion significant with AU tests for the masked-unpartitioned and [27] with an extended majority-rule (MRE) consensus tree masked-partitioned approach. Our split network analyses similarly criterion and a cutoff value of 0.01. Thus, the number of 5,000 corroborate this scenario. This is also congruent to earlier bootstrap replicates chosen for our ML analyses had been molecular studies [3,4], but contradicts the results of Rabeling et sufficient for all of our data sets. al. [1]. Based on our re-analyses of the respective data set [1] and Our partitioned ML analysis of the masked data set clearly other molecular studies [3–5,29,30], we suggest that, at present, it outperformed the masked-unpartitioned data set in terms of seems unlikely that Martialinae are the sister group to all other likelihood scores (masked-partitioned: ln = 249230.716; masked- recent ant subfamilies. unpartitioned: ln = 252002.229). The placement of Martialinae suggested by Rabeling et al. [1] could be due to inferior sequence alignments or confounding Phylogenetic relationships effects of randomized alignment sections. The MAFFT-L-ins-i Placement of Leptanillinae and Martialinae. All ML and algorithm applied in our study was shown to be one of the most Bayesian topologies suggested a clade including Leptanillinae + all accurate available alignment algorithms, and can be considered to remaining ant subfamilies with maximum support (Figures 1, 2, 3, be the best choice for sequence alignments [31,32]. Still, 739 Table 1, and Figures S2, S3, S4, S5, S6, S7). Martialinae always split alignment positions were identified by ALISCORE as potentially off as a second branch and form a clade with poneroids and randomised and therefore excluded. ALISCORE and subsequent monophyletic formicoids. Applying an approximately unbiased test alignment masking increased the signal-to-noise ratio within the (AU test) [28] for all ML topologies, the Null hypothesis (H0) data, but influenced our tree topologies only marginally. However, assumes that either Leptanillinae as a sister group of remaining a positive effect of the masking approach is clearly shown by a Formicidae and Martialinae as second branch in the ant tree of life strong decrease of contradictory signal within the masked or vice versa, are not significantly different. While H0 was not alignment, especially for deeper splits (Figure S1). Partitioning of significantly rejected for our unmasked data set (p = 0.120), both ML the masked data set leads to an increased likelihood score, and

PLoS ONE | www.plosone.org 2 June 2011 | Volume 6 | Issue 6 | e21031 Improved Analyses for the Ant Tree of Life

Figure 1. ML topology inferred from the unmasked, unpartitioned data set. Schematised ML topology with branch lengths inferred from the unmasked supermatrix (best ML tree, majority rule, 5,000 bootstrap replicates). Quotation marks indicate non-monophyly. doi:10.1371/journal.pone.0021031.g001 higher node resolution within formicoids. Martialinae are again it should be considered that the subfamily Martialinae was just resolved as the second branch (cf. Figures 1 2, 3, Table 1, and discovered in 2008. Therefore, Moreau (2009) [29] combined data Figures S2, S3, S4, S5, S6, S7) avoiding possible artifacts due to sets of Brady et al. (2006) [3], Moreau et al. (2006) [4], and noise. Rabeling et al. (2008) [1] to a supermatrix in which the Discrepancies between our results and the results of Rabeling relationship of Leptanillinae and Martialinae was unresolved. et al. [1] could further be explained by an insufficient number of Our analyses showed that an exclusion of randomised sections boostrap replicates (ML approach) and an insufficient number of improved the resolution between Ponerinae and the formicoids Bayesian generations. They conducted 500 bootstrap replicates for (Figure 2, 3 and Figures S1, S3, S4, S6, S7). Alignment masking the ML approach [1] versus 5,000 bootstrap replicates in our study. led to a placement of Ponerinae next to formicoids (Table 1). Pattengale et al. (2010) [27] showed in a recent study on Discrepancies between low bs and high bpp support values seem to ‘bootstopping’ that the number of bootstrap replicates for accurate confirm typical observations considering Bayesian analyses [10– confidence values is strongly dependent on the data set. In testing 13]. The relationships between the Amblyoponinae, Agroecomyr- the performance and accuracy of bootstrap criteria on real DNA mecinae, Paraponerinae, and Proceratiinae remain unresolved in alignments, they showed that a range of 100 – 500 bootstrap most of our topologies. Only the Bayesian topology of the masked- replicates is usually sufficient. Still, in some cases a much higher partitioned data set show monophyletic Amblyoponinae with weak number of up to 1,200 replicates was necessary to deliver support support (Tab. 1). Thereby, Amblyoponinae branch off as a third values that are equally robust as those in the reference tree with split (0.84 bpp) within the ant tree of life. The monophyly of 10,000 replicates. Most differences between reference and ‘boot- Amblyoponinae has been favoured by earlier studies [3–5,29]. stopped’ topologies occurred on poorly supported branches Therefore, we conclude that more genes are necessary to robustly (,75% bs). Since the bootstrap support in the ML tree of resolve an amblyoponine clade as well as relationships between Rabeling et al. [1] for a clade Martialinae+remaining ants is only Amblyoponinae, Agroecomyrmecinae, Paraponerinae, and Pro- 76.2%, 500 replicates might have been insufficient. In contrast, ceratiinae. All our topologies highly support a dorylomorph clade. our support values derived from 5,000 bootstrap replicates are Our unmasked and masked-partitioned topology and both evaluated and confirmed by a posteriori ‘bootstop tests’ (see results). Bayesian topologies derived from our masked approaches As mentioned above, single data sets of earlier studies [3,4] corroborate a placement of the dorylomorphs next to the propose Leptanillinae as a sister lineage to all other ants. However, remaining formicoids. This hypothesis stands in concordance with

PLoS ONE | www.plosone.org 3 June 2011 | Volume 6 | Issue 6 | e21031 Improved Analyses for the Ant Tree of Life

Figure 2. ML topology inferred from the masked-unpartitioned data set. Schematised ML topologies with branch lengths inferred from the masked supermatrix. Best ML tree of the masked-unpartitioned analysis (739 positions excluded from the unmasked alignment), majority rule, 5,000 bootstrap replicates. Quotation marks indicate non-monophyly. doi:10.1371/journal.pone.0021031.g002 other studies [1,3,4]. Finally, the non-monophyly of cerapachyines to support a robust ant tree (cf. Figure S1). For a deeper insight within the dorylomorphs is consistent with these studies. into subfamily relationships, multi-gene analyses of genomic/EST Compared with Brady et al. 2006 [3], the inclusion of data and a more exhaustive taxon sampling combined with Martialinae reduce the branch lengths for leptanillines and improved phylogenetic approaches seem indispensable. formicids, although the branch separating ants from the aculeate outgroup Hymenoptera still remains relatively long. However, Materials and Methods with current methods and the available data, it is not possible to assess putative long branch artifacts like discussed in Brady et al. Data set 2006 [3]. It is possible that new molecular sequence data might We used molecular data previously published by Rabeling et al. ‘improve’ the current ant tree of life. It is possible that a data set (2008) [1]. In accordance to [1], we used the data matrix of Brady with most signal coming from rRNA genes might not be sufficient et al.(2006) [3] kindly provided by S. Brady. We added respective

PLoS ONE | www.plosone.org 4 June 2011 | Volume 6 | Issue 6 | e21031 Improved Analyses for the Ant Tree of Life

Figure 3. ML topology inferred from the masked-partitioned data set. Schematised ML topologies with branch lengths inferred from the masked supermatrix. Best ML tree, of the masked-partitioned analysis (739 positions excluded from the unmasked alignment+one bp to correct the reading frame), majority rule, 5,000 bootstrap replicates. Quotation marks indicate non-monophyly. doi:10.1371/journal.pone.0021031.g003 sequences of Martialis heureka [1] from GenBank (http://www.ncbi. methods in benchmark tests [31,32]. Each of the three sequence nlm.nih.gov/). The data set comprised three genes of 152 taxa alignments (18S, 28S, and EF1aF2) was screened for randomised subdivided into 21 ant subfamilies and 11 outgroup taxa. sections with ALISCORE [17] using all possible pairwise Sequence data included elongation factor 1-alpha F2 (EF1aF2, comparisons and a window size w = 6. Within ALISCORE, gaps nuclear protein coding gene), 18S rRNA and 28S rRNA (nuclear were treated as ambiguous characters. Randomised sections (28S ribosomal genes). rRNA: 725 base positions (bp); 18S rRNA: 14 bp) were excluded with ALICUT [34]. In the EF1aF2 alignment, no randomised Alignment positions were detected. Single genes were concatenated using Single genes were aligned separately using the local L-ins-i FASconCAT version 1.0 [35]. The concatenated supermatrix of algorithm of MAFFT version 6.717 [33]. The L-ins-i algorithm is the masked approach included 4,315 characters while the an iterative progressive algorithm which outperformed other unmasked supermatrix comprised 5,054 characters. All alignments

PLoS ONE | www.plosone.org 5 June 2011 | Volume 6 | Issue 6 | e21031 Improved Analyses for the Ant Tree of Life

Table 1. Selected clades with posterior probability and bootstrap support values.

Bayes posterior probabilities [bpp] ML bootstrap support [bs]

unmasked masked masked-part. unmasked masked masked-part.

Clade 1 1 1 1 100 100 100 Clade 2 1 1 1 90 93 93 poneroids 0.86 – – – – – Amblyoponinae – – 0.77 – – – (Ponerinae,formicoids) – 0.97 1 – 57 68 formicoids 1 1 1 100 100 100 dorylomorphs 1 1 1 100 100 100

Selected clades with bayesian posterior probability [bpp] and bootstrap support [bs] values recovered in our Bayesian (Bayes) and Maximum Likelihood (ML) topologies. Clade 1 (Leptanillinae,(Martialinae, remaining ants)) and (Martialinae(poneroid/formicoid clade)) are resolved in all Bayesian and ML topologies. Poneroids are not monophyletic with the exception of the unmasked, Bayesian topology (weakly supported). Amblyoponinae are only monophyletic within the Bayesian masked- partitioned topology. A clade (Ponerinae,formicoids) with a subsequent paraphyly of poneroids, is suggested by all masked topologies with high Bayesian posterior probability (bpp) but low bootstrap (bs) support. Dorylomorphs are monophyletic with exception of the masked-unpartitioned ML topology.

(fasta format) and the respective character partitions are provided (unmasked, masked-unpartitioned and masked-partitioned). We in Information S1, S2, S3, S4 and are freely available from http:// compared alternative tree topologies by performing an AU test www.zfmk.de. [28] for each data set. Therefore, we optimised branch lengths for alternative topologies. Subsequently, we calculated per site log Phylogenetic reconstructions Likelihood scores using RAxML 7.2.6. AU tests were performed Split networks. We computed NeighbourNetworks [24–26] with CONSEL [40], version v0.1i. with SplitsTree 4.10 [25] to visualise the data structure of the Bayesian Analyses. Bayesian phylogenies were calculated unmasked and masked alignments. NeighborNetworks were calcu- using MrBayes [41,42] for three data sets also used in our ML lated applying uncorrected p-distances for the unmasked analyses. Topologies were inferred from (i) the unmasked alignment and the masked alignment used for the masked- superalignment (ii) the masked superalignment, non-partitioned partitioned analyses. NeighborNetwork graphs give an indication and (iii) the masked superalignment with four partitions according of noise, signal-like patterns and conflicts within a multiple to [1] and our ML analyses. Similar to Rabeling et al., we used sequence alignments. MrBayes v3.2 (an unreleased version of MrBayes; the source code Maximum Likelihood Analyses. We estimated a Maximum was downloaded from the current version system in January, Likelihood (ML) topology for the unmasked supermatrix and the 2011). Convergence of parameters of the Bayesian analyses was masked supermatrix in non-partitioned analyses with RAxML assessed with the software Tracer v1.5 [43]. [36] using RAxMLHPC-PTHREADS [37], version 7.2.6. A We chose the sequence evolution model GTR+C for all three third topology was reconstructed from the masked supermatrix data sets (i) – (iii) for accuracy of comparison with our ML with four partitions according to the setup described for the analyses. Parameters of the model (i.e., base frequencies, Bayesian analyses in Rabeling et al. (2008) [1] with the transition/transversion ratio, and rate variation shape parameter) RAxMLHPC-HYBRID [38], version 7.2.6. The first partition were unlinked across partitions. According to Rabeling et al., included the 18S, the second partition the 28S. The third Metropolis coupling was used with eight chains per analysis and a partition comprised the 1st and 2nd codon position of EF1aF2, temperature increment of 0.05 [1]. For analysis (i) and (ii) we ran the fourth partition included the 3rd codon position of EF1aF2. 30 million generations with a sample frequency of 200. For We identified the correct reading frame and excluded the first analysis (iii) we ran 28,130,500 generations with a sample position of the EF1aF2-alignment. Therefore, the EF1aF2- frequency of 100. After checking all analyses for parameter alignment was 1 bp shorter (516 bp) than that described in convergence in Tracer v1.5, we discarded a burn-in of 10% for Rabeling et al. (2008) [1]. each analysis. After discarding the burn-in, majority rule We conducted rapid bootstrap analyses and a thorough search consensus trees with posterior probabilities were calculated from for the best ML tree using GTR+a with 5,000 bootstrap replicates. all sampled trees within MrBayes. All analyses were performed on We evaluated the number of necessary bootstrap replicates a HPC LINUX clusters of the ZFMK, Bonn, Germany. Trees were posteriori for each data set according to the bootstop criteria based on edited with the software TreeGraph 2 [39]. the Weighted Robinson-Foulds (WRF) distance criterion [27] using RAxML 7.2.6 for the extended majority-rule (MRE) consensus tree Supporting Information criterion. We chose a cutoff value of 0.01 to ensure a sufficient Figure S1 NeighborNet graphs with uncorrected p number of bootstrap replicates. In final trees, clades with a bootstrap distances inferred with Splitstree version 4.10 from the support (bs) below 50% were considered unresolved. All analyses unmasked and masked alignment. were performed on HPC LINUX clusters of the ZFMK, Bonn, (PDF) Germany. Trees were edited with the software TreeGraph 2 [39]. To test alternative placements of Martialinae and Leptanillinae Figure S2 RAxML-phylogram (majority rule) inferred as suggested by Rabeling et al. (2008) [1], we exchanged the from the unmasked alignment. position of Martialinae and Leptanillinae in our best trees (PDF)

PLoS ONE | www.plosone.org 6 June 2011 | Volume 6 | Issue 6 | e21031 Improved Analyses for the Ant Tree of Life

Figure S3 RAxML-phylogram (majority rule) inferred Information S2 Masked alignment in fasta format used from the masked-unpartitioned approach. for the masked-unpartitioned analyses. (PDF) (PHY) Figure S4 RAxML-phylogram (majority rule) inferred Information S3 Masked alignment in fasta format used from the masked-partitioned approach. (refer to for the masked-partitioned analyses. Figure 2 and 3 in the manuscript). (PHY) (PDF) Information S4 Character partition file (plain text Figure S5 Bayesian-phylogram (majority rule consen- format) for the masked alignment used for the sus tree) inferred from the unmasked alignment masked-partitioned analyses. (28,130,500 generations, samplefrequency 100, burn-in: (PHY) 10% discarded). (PDF) Acknowledgments Figure S6 Bayesian-phylogram (majority rule consen- We are grateful to Sea´n G. Brady for providing the alignment of data sus tree) inferred from the masked-unpartitioned published in 2006. We thank Julia Schwarzer and Carola Greve for all of approach (30 million generations, sample frequency their help with Bayesian analyses. Thanks to Jonas Astrin, Christoph 200, burn-in: 10% discarded). Mayer and Johann-Wolfgang Wa¨gele for their helpful comments, and Eva (PDF) Wiesel for her help with the figures. Special thanks go to Birthe Thormann and Lily Wescott for proofreading. Figure S7 Bayesian-phylogram (majority rule consensus tree) inferred from the masked-partitioned approach. (30 million generations, sample frequency 200, Author Contributions burn-in: 10% discarded). Conceived and designed the experiments: PK FHG. Analyzed the data: (PDF) PK KM. Wrote the paper: PK KM BM FHG. Designed the setup and performed all analyses: PK KM. Designed the figures: KM. Read and Information S1 Unmasked alignment in fasta format. approved the final manuscript: PK FHG BM KM. (PHY)

References 1. Rabeling C, Brown J, Verhaagh M (2010) Newly discovered sister lineage sheds 19. Schwarzer J, Misof B, Tautz D, Schliewen UK (2009) The root of the East light on early ant evolution. Proc Natl Acad Sci U S A 105: 14913–14917. African cichlid radiations. BMC Evol Biol 9: 186. 2. Bolton B (1990) The higher classification of the ant subfamily Leptanillinae 20. Simon S, Strauss S, von Haeseler A, Hadrys H (2009) A phylogenomic approach (Hymenoptera: Formicidae). Syst Ent 15: 267–282. to resolve the basal pterygote divergence. Mol Biol Evol 26: 2719–2730. 3. Brady S, Schultz T, Fisher B, Ward P (2006) Evaluating alternative hypotheses 21. Meusemann K, von Reumont BM, Simon S, Roeding F, Ku¨ck P, et al. (2010) A for the early evolution and diversification of ants. Proc Natl Acad Sci U S A 103: phylogenomic approach to resolve the arthropod tree of life. Mol Biol Evol 27: 18172–18177. 2451–2464. 4. Moreau C, Bell C, Vila R, Archibad S, Pierce N (2006) Phylogeny of the ants: 22. Murienne J, Edgecombe G, Giribet G (2010) Including secondary structure, diversification in the age of angiosperms. Science 312: 101–104. fossils and molecular dating in the centipede tree of life. Mol Phylogenet Evol 57: 5. Ouellette G, Fisher B, Girman D (2006) Molecular systematics of basal 301–313. subfamilies of ants using 28S rRNA (Hymenoptera: Formicidae). Mol 23. Dinapoli A, Zinssmeister C, Klussmann-Kolb A (2011) New insights into the Phylogenet Evol 40: 359–369. phylogeny of the Pyramidellidae (Gastropoda). J Mollus Stud 77: 1–7. 6. Baroni Urbani C, Bolton B, Ward PS (1992) The internal phylogeny of ants 24. Bryant D, Moulton V (2004) Neighbor-Net: An agglomerative method for the (Hymenoptera: Formicidae). Syst Ent 17: 301–329. construction of phylogenetic networks. Mol Biol Evol 21: 255–265. 7. Grimaldi D, Agosti D, Carpenter J (1997) New and rediscovered primitive ants 25. Huson DH, Bryant D (2006) Application of phylogenetic networks in (Hymenoptera: Formicidae) in Cretaceous amber from New Jersey, and their evolutionary studies. Mol Biol Evol 23: 254–267. phylogenetic relationship. Am Mus Novit 3208: 1–43. 26. Bryant D, Moulton V, Spillner A (2007) Consistency of the neighbor-net 8. Grimaldi D, Agosti D (2000) A formicine in New Jersey Cretaceous amber algorithm. Algorithms Mol Biol 2: 8. (Hymenoptera: Formicidae) and early evolution of the ants. Proc Natl Acad 27. Pattengale N, Alipour M, Bininda-Emonds ORP, Moret BME, Gottlieb EJ, Sci U S A 97: 13678–13683. et al. (2010) How many bootstrap replicates are necessary? J Comput Biol 17: 9. Wilson EO, Ho¨lldobler B (2005) The rise of the ants: a phylogenetic and 337–354. ecological explanation. Proc Natl Acad Sci U S A 102: 7411–7414. 28. Shimodaira H (2002) An approximately unbiased test of phylogenetic tree 10. Suzuki Y, Glazko GV, Nei M (2002) Overcredibility of molecular phylogenies selection. Syst Biol 51: 492–508. obtained by Bayesian phylogenetics. Proc Natl Acad Sci U S A 99: 29. Moreau C (2009) Inferring ant evolution in the age of molecular data 16138–16143. (Hymenoptera: Formicidae). Myrmecol News 12: 201–210. 11. Erixon P, Svennblad B, Britton T, Oxelman B (2003) Reliability of Bayesian 30. Ward PS (2010) Taxonomy, phylogenetics, and evolution. In: Lach L, Parr CL, posterior probabilities and bootstrap frequencies in phylogenetics. Syst Biol 52: Abbott K, editors, Ant Ecology, Oxford: Oxford University Press, chapter 1. pp 665–673. 3–17. 12. Huelsenbeck J, Larget B, Miller R, Ronquist F (2002) Potential applications and 31. Nuin P, Wang Z, Elisabeth R (2006) The accuracy of several multiple sequence pitfalls of Bayesian inference of phylogeny. Syst Biol 51: 673–688. alignment programs for proteins. BMC Bioinformatics 7: 471. 13. Douady C, Delsuc F, Boucher Y, Doolittle W, Douzery E (2003) Comparison of 32. Golubchik T, Wise M, Easteal S, Jermiin L (2007) Mind the gaps: evidence of Bayesian and Maximum Likelihood bootstrap measures of phylogenetic bias in estimates of multiple sequence alignments. Mol Biol Evol 24: 2433–2442. reliability. Mol Biol Evol 20: 248–254. 33. Katoh K, Toh H (2008) Recent developments in the MAFFT multiple sequence 14. Castresana J (2000) Selection of conserved blocks from multiple alignments for alignment program. Brief Bioinform 9: 286–298. their use in phylogenetic analysis. Mol Biol Evol 17: 540–552. 34. Ku¨ck P (2009) ALICUT: a Perlscript which cuts ALISCORE identified RSS. 15. Dress A, Flamm C, Fritzsch G, Gru¨newald S, Kruspe M, et al. (2008) Noisy: Department of Bioinformatics, Zoologisches Forschungsmuseum A. Koenig Identification of problematic columns in multiple sequence alignments. (ZFMK), Bonn, Germany, version 2.0 edition. Algorithms Mol Biol 3: 7. 35. Ku¨ck P, Meusemann K (2010) FASconCAT: Convenient handling of data 16. Hartmann S, Vision TJ (2008) Using ESTs for phylogenomics: Can one accurately matrices. Mol Phylogenet Evol 56: 1115–1118. infer a phylogenetic tree from a gappy alignment? BMC Evol Biol 8: 95: S13. 36. Stamatakis A (2006) RAxML-VI-HPC: maximum likelihood-based phylogenetic 17. Misof B, Misof K (2009) A Monte Carlo approach successfully identifies analyses with thousands of taxa and mixed models. Bioinformatics 22: randomness in multiple sequence alignments: a more objective means of data 2688–2690. exclusion. Syst Biol 58: 21–34. 37. Ott M, Zola J, Aluru S, Stamatakis A (2007) Large-scale maximum likelihood- 18. Ku¨ck P, Meusemann K, Dambach J, Thormann B, von Reumont B, et al. (2010) based phylogenetic analysis on the IBM BlueGene/L. In: Proceedings of ACM/ Parametric and non-parametric masking of randomness in sequence alignments IEEE Supercomputing conference 2007 Reno NV., ACM, New York, NY, can be improved and leads to better resolved trees. Front Zool 7: 10. USA.

PLoS ONE | www.plosone.org 7 June 2011 | Volume 6 | Issue 6 | e21031 Improved Analyses for the Ant Tree of Life

38. Pratas F, Trancoso P, Stamatakis A, Sousa L (2009) Fine-grain parallelism using 41. Huelsenbeck J, Ronquist F (2001) MrBayes: Bayesian inference of phylogenetic multi-core, Cell/BE, and GPU systems: Accelerating the phylogenetic likelihood trees. Bioinformatics 17: 754–755. function. In: Proceedings of ICPP 2009 Vienna, Austria, IEEE Computer 42. Ronquist F, Huelsenbeck J (2003) MrBayes 3: Bayesian phylogenetic inference Society, Los Alamitos, CA, USA. under mixed models. Bioinformatics 19: 1572–1574. 39. Sto¨ver BC, Mu¨ller KF (2010) TreeGraph 2: combining and visualizing evidence 43. Rambaut A, J DA Tracer. Available: http://beast.bio.ed.ac.uk/software/ from different phylogenetic analyses. BMC Bioinformatics 11: 7. tracer/, version 1.5 edition. 40. Shimodaira H, Hasegawa M (2001) CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17: 12461247.

PLoS ONE | www.plosone.org 8 June 2011 | Volume 6 | Issue 6 | e21031 Pseudo- a myrmecinae

Myrme- dolicho- ciinae Martialis deromorphs Proceratiinae Formicinae 'Amblyoponinae'

Agroecomyrmecinae Paraponerinae

ectaheteromorphs Ponerinae

Leptanillidae

Myrmicinae

outgroup taxa Martialinae Leptanillinae ponieroids formicoids outgroup taxa dorylomorphs

0.01

outgroup taxa

Martialis Proceratiinae Myrmicinae Leptanillidae

'Amblyoponinae' Agroecomyrmecinae

Paraponerinae ectaheteromorphs

Ponerinae

Myrme- Formicinae ciinae

Pseudo- dolicho- myrmecinae deromorphs outgroup taxa Martialinae Leptanillinae ponieroids 0.01 dorylomorphs formicoids

Figure S1: Neighbornet graphs with uncorrected p distances inferred with Splitstree version 4.10. a: Split network based on the unmasked alignment. b: Split network based on the masked alignment which was used for the masked-partitioned analyses. Pristocera MAD01 100 Mischocyttarus flavitarsis Metapolybia cingulata Aglyptacros cf. sulcatus Scolia verticalis Chyphotes mellipes Sapyga pumila outgroup taxa Aporus niger Dasymutilla aureola 96 Apis mellifera Chalybion californicum

81 100 Protanilla JAP01 100 Leptanilla GRE01 Leptanillinae Leptanilla RSA01 Martialis heureka Tatuidris ECU01 Martialinae 99 Onychomyrmex hedleyi 99 Concoctio concenta Prionopelta MAD01

84 Amblyopone pallipes 100 92 100 Amblyopone mutica Mystrium mysticum Adetomyrma MAD02 Apomyrma stygia Paraponera clavata 61 Discothyrea MAD07 97 Probolomyrmex tani 100 Proceratium MAD08 Proceratium stictum 'poneroids' 91 100 Platythyrea punctata Platythyrea mocquerysi Anochetus madagascarensis Leptogenys diminuta 99 Pachycondyla sikorae 100 Odontomachus coquereli

53 Odontoponera transversa

88 Centromyrmex sellaris 85 Psalidomyrmex procerus 90 93 Loboponera politula 100 Plectroctena ugandensis 98 Hypoponera sakalava Hypoponera opacior 93 Thaumatomyrmex atrox

94 Simopelta cf. pergandei Cerapachys sexspinus 59 Cerapachys augustae Cerapachys larvatus Simopone marleyi 60 Sphinctomyrmex steinheili 92 Cylindromyrmex striatus 100 89 Acanthostichus kirbyi 100 Leptanilloides nomada Leptanilloides mckennae 100 Dorylus laevigatus 73 Dorylus helvolus Aenictogiton ZAM02 65 100 Aenictus ceylonicus 72 Aenictus eugenii

100 Cheliomyrmex cf. morosus 100 Neivamyrmex nigrescens Eciton vagans

100 Tetraponera punctulata 88 Tetraponera rufonigra 99 Myrcidris epicharis Pseudomyrmex gracilis Aneuretus simoni Dolichoderus scabridus 94 100 Liometopum occidentale 99 Liometopum apiculatum 100 99 Tapinoma sessile 100 Technomyrmex difficilis Technomyrmex MAD05 91 100 94 Leptomyrmex AUS01 Leptomyrmex erythrocephalus 55 Dorymyrmex bicolor Forelius pruinosus 66 Papyrius nitidus Azteca ovaticeps

53 Linepithema humile Anonychomyrma gilberti 67 Turneria bidentata 79 Philidris cordatus 100 Myrmecia pyriformis Nothomyrmecia macrops 100 Myrmecocystus flaviceps Lasius californicus 100 Brachymyrmex depilis Myrmelachista JTL01 Oecophylla smaragdina 79 Notostigma carazzii 100 Polyergus breviceps Formica moki 100 Polyrhachis Cyrto01 93 Polyrhachis Hagio01

72 Calomyrmex albertisi 94 Camponotus hyatti 60 95 72 Camponotus maritimus

100 88 Camponotus conithorax Camponotus BCA01 formicoids Opisthopsis respiciens Anoplolepis gracilipes Myrmoteras iriodum Notoncus capitatus Acropyga acutiventris 71 Pseudolasius australis

84 Paratrechina hystrix Prenolepis albimaculata Prenolepis imparis 100 Myrmica tahoensis 100 Myrmica striolagaster 83 Manica bradleyi Pogonomyrmex subdentatus Orectognathus versicolor Daceton armigerum 55 Microdaceton tibialis Eurhopalothrix bolaui Acanthognathus ocellatus Myrmicocrypta cf infuscata Apterostigma auriculatum 100 Acromyrmex versicolor Trachymyrmex arizonensis Wasmannia auropunctata 100 Procryptocerus scabriusculus 100 Pyramica hoplites Strumigenys dicomas 99 Pheidole clydei Pheidole hyatti Basiceros manni 53 Pilotrochus besmerus 90 Aphaenogaster albisetosa Messor andrei

62 Stenamma dyscheres 99 Aphaenogaster swammerdami 100 Aphaenogaster occidentalis 95 Messor denticornis 100 Solenopsis molesta 92 Solenopsis xyloni 78 Myrmicaria exigua Monomorium ergatogyna Vollenhovia emeryi 100 Tetramorium validiusculum Tetramorium caespitum Mayriella ebbei Xenomyrmex floridanus Eutetramorium mocquerysi Myrmecina graminicola Crematogaster emeryana Cardiocondyla mauritanica Leptothorax muscorum complex Nesomyrmex echinatinodis 99 Meranoplus cf. radamae Cataulacus MAD02 Terataner MAD02 Pheidologeton affinis Temnothorax rugatulus Metapone madagascarica Rhopalomastix rothneyi 100 Acanthoponera minor Heteroponera panamensis 85 85 Gnamptogenys striatula 100 Typhlomyrmex rogenhoferi 86 Ectatomma opaciventre Rhytidoponera chalybaea

0.1

Figure S2: Maximum likelihood (majority rule) inferred from the unmasked, unpartitioned data set with 5,000 bootstrap replicates (-f a; GTR + GAMMA, see method section). The tree was rooted with Pristocera. Pristocera MAD01 100 Mischocyttarus flavitarsis Metapolybia cingulata Aglyptacros cf. sulcatus Scolia verticalis Chyphotes mellipes 55 Sapyga pumila outgroup taxa Dasymutilla aureola Aporus niger 94 Chalybion californicum

79 Apis mellifera 100 Protanilla JAP01 100 Leptanilla GRE01 Leptanillinae Leptanilla RSA01 Martialis heureka Tatuidris ECU01 Martialinae

98 Onychomyrmex hedleyi 99 Concoctio concenta 100 Prionopelta MAD01

89 Amblyopone pallipes 91 99 Amblyopone mutica Mystrium mysticum Adetomyrma MAD02 Apomyrma stygia Paraponera clavata 93 Discothyrea MAD07 96 Probolomyrmex tani 100 Proceratium MAD08 Proceratium stictum 100 Platythyrea punctata Platythyrea mocquerysi Anochetus madagascarensis 'poneroids' 100 100 Leptogenys diminuta Pachycondyla sikorae 62 Odontoponera transversa

100 Odontomachus coquereli 94 92 Thaumatomyrmex atrox Simopelta cf. pergandei 51 78 Centromyrmex sellaris 73 Psalidomyrmex procerus 93 87 Loboponera politula Plectroctena ugandensis 100 Hypoponera sakalava Hypoponera opacior Cerapachys sexspinus Cerapachys augustae Cerapachys larvatus Simopone marleyi Sphinctomyrmex steinheili 50 89 Cylindromyrmex striatus

100 Acanthostichus kirbyi 100 Leptanilloides nomada Leptanilloides mckennae 100 Dorylus helvolus 59 Dorylus laevigatus Aenictogiton ZAM02 100 Aenictus ceylonicus

67 Aenictus eugenii

57 100 Cheliomyrmex cf. morosus 100 Neivamyrmex nigrescens Eciton vagans

100 Tetraponera punctulata 93 Tetraponera rufonigra 97 Myrcidris epicharis Pseudomyrmex gracilis Aneuretus simoni Dolichoderus scabridus 100 71 Liometopum apiculatum 98 Liometopum occidentale 99 Tapinoma sessile 100 100 Technomyrmex difficilis Technomyrmex MAD05 93 Leptomyrmex AUS01 Leptomyrmex erythrocephalus 65 Dorymyrmex bicolor Forelius pruinosus 70 Papyrius nitidus Azteca ovaticeps

58 Linepithema humile Anonychomyrma gilberti 64 Turneria bidentata 65 Philidris cordatus 100 Myrmecia pyriformis Nothomyrmecia macrops 99 Lasius californicus 100 Myrmecocystus flaviceps 99 Brachymyrmex depilis Myrmelachista JTL01 formicoids Oecophylla smaragdina

77 Notostigma carazzii 100 Polyergus breviceps Formica moki 100 Polyrhachis Cyrto01

63 Polyrhachis Hagio01 Calomyrmex albertisi

62 94 Camponotus hyatti 91 95 Camponotus maritimus 100 86 Camponotus BCA01 Camponotus conithorax Opisthopsis respiciens Anoplolepis gracilipes Myrmoteras iriodum Notoncus capitatus Acropyga acutiventris Pseudolasius australis Paratrechina hystrix 85 Prenolepis imparis 55 Prenolepis albimaculata 100 Myrmica tahoensis 100 Myrmica striolagaster 73 Manica bradleyi Pogonomyrmex subdentatus Orectognathus versicolor Daceton armigerum Microdaceton tibialis Eurhopalothrix bolaui Acanthognathus ocellatus Myrmicocrypta cf. infuscata Apterostigma auriculatum 51 100 Acromyrmex versicolor Trachymyrmex arizonensis Wasmannia auropunctata 100 Procryptocerus scabriusculus 100 Pyramica hoplites Strumigenys dicomas 96 Pheidole clydei Pheidole hyatti Basiceros manni Pilotrochus besmerus 96 Aphaenogaster albisetosa Messor andrei Stenamma dyscheres

94 Aphaenogaster swammerdami Aphaenogaster occidentalis 97 100 Messor denticornis 100 Solenopsis molesta 86 Solenopsis xyloni 83 Myrmicaria exigua Monomorium ergatogyna Vollenhovia emeryi 73 100 Tetramorium validiusculum Tetramorium caespitum Mayriella ebbei Xenomyrmex floridanus Eutetramorium mocquerysi Myrmecina graminicola Crematogaster emeryana Cardiocondyla mauritanica Leptothorax muscorum complex Nesomyrmex echinatinodis 75 Meranoplus cf. radamae Cataulacus MAD02 Terataner MAD02 Pheidologeton affinis Temnothorax rugatulus Metapone madagascarica Rhopalomastix rothneyi 100 Acanthoponera minor Heteroponera panamensis 91 88 Typhlomyrmex rogenhoferi 100 Gnamptogenys striatula 96 Ectatomma opaciventre Rhytidoponera chalybaea

0.1 Figure S3 : Maximum likelihood (majority rule) inferred from the masked, unpartitioned data set with 5,000 bootstrap replicates (-f a, GTR + GAMMA, see method section). The tree was rooted with Pristocera. Pristocera MAD01 100 Mischocyttarus flavitarsis Metapolybia cingulata Aglyptacros cf. sulcatus Scolia verticalis Chyphotes mellipes outgroup taxa 72 Sapyga pumila Dasymutilla aureola Aporus niger 87 Chalybion californicum 62 Apis mellifera

100 Protanilla JAP01 100 Leptanilla GRE01 Leptanillinae Leptanilla RSA01 Martialis heureka Tatuidris ECU01 Martialinae

100 Onychomyrmex hedleyi 100 Concoctio concenta 100 Prionopelta MAD01 Amblyopone pallipes 100 Amblyopone mutica Mystrium mysticum Adetomyrma MAD02 Apomyrma stygia 93 Paraponera clavata

98 Discothyrea MAD07 80 Probolomyrmex tani 100 Proceratium MAD08 Proceratium stictum 100 Platythyrea punctata Platythyrea mocquerysi Anochetus madagascarensis 'poneroids' 100 100 Leptogenys diminuta Pachycondyla sikorae Odontoponera transversa 91 55 100 74 Odontomachus coquereli 98 Thaumatomyrmex atrox Simopelta cf. pergandei 84 91 Centromyrmex sellaris Psalidomyrmex procerus

98 89 Loboponera politula 96 Plectroctena ugandensis 100 Hypoponera sakalava Hypoponera opacior Cerapachys sexspinus Cerapachys augustae Cerapachys larvatus Simopone marleyi 100 Dorylus laevigatus 62 Dorylus helvolus Aenictogiton ZAM02 100 Aenictus ceylonicus 100 73 Aenictus eugenii

100 Cheliomyrmex cf. morosus 100 Neivamyrmex nigrescens

68 Eciton vagans 66 Acanthostichus kirbyi Cylindromyrmex striatus 100 Leptanilloides mckennae Leptanilloides nomada Sphinctomyrmex steinheili

100 Tetraponera punctulata 99 Tetraponera rufonigra 98 Myrcidris epicharis 57 Pseudomyrmex gracilis Aneuretus simoni Dolichoderus scabridus 81 100 Liometopum occidentale 95 Liometopum apiculatum 100 97 Tapinoma sessile 100 Technomyrmex difficilis

94 Technomyrmex MAD05 Leptomyrmex AUS01 70 Leptomyrmex erythrocephalus Dorymyrmex bicolor 59 Forelius pruinosus 77 Papyrius nitidus 100 68 Azteca ovaticeps

58 Linepithema humile Anonychomyrma gilberti

69 Turneria bidentata 64 Philidris cordatus 100 Myrmecia pyriformis Nothomyrmecia macrops 100 Lasius californicus Myrmecocystus flaviceps 98 Brachymyrmex depilis Myrmelachista JTL01 Oecophylla smaragdina

86 Notostigma carazzii 100 Polyergus breviceps Formica moki 100 Polyrhachis Cyrto01 formicoids 64 Polyrhachis Hagio01 Calomyrmex albertisi

96 70 Camponotus hyatti 93 Camponotus maritimus 84 100 Camponotus BCA01 82 Camponotus conithorax Opisthopsis respiciens Anoplolepis gracilipes Myrmoteras iriodum 63 Notoncus capitatus Acropyga acutiventris Pseudolasius australis Paratrechina hystrix 88 Prenolepis imparis Prenolepis albimaculata 100 Myrmica tahoensis 100 Myrmica striolagaster Manica bradleyi Pogonomyrmex subdentatus Orectognathus versicolor Daceton armigerum Microdaceton tibialis Eurhopalothrix bolaui Acanthognathus ocellatus Myrmicocrypta cf. infuscata Apterostigma auriculatum 100 Acromyrmex versicolor Trachymyrmex arizonensis

100 Wasmannia auropunctata Procryptocerus scabriusculus 100 Pyramica hoplites Strumigenys dicomas 99 Pheidole clydei Pheidole hyatti Basiceros manni Pilotrochus besmerus 98 Aphaenogaster albisetosa Messor andrei Stenamma dyscheres

79 Aphaenogaster swammerdami 98 Aphaenogaster occidentalis 97 100 Messor denticornis 100 Solenopsis molesta 90 Solenopsis xyloni 60 Myrmicaria exigua Monomorium ergatogyna Vollenhovia emeryi 100 Tetramorium validiusculum 67 Tetramorium caespitum Mayriella ebbei Xenomyrmex floridanus Eutetramorium mocquerysi Myrmecina graminicola Crematogaster emeryana Cardiocondyla mauritanica 65 Leptothorax muscorum complex Nesomyrmex echinatinodis 88 Meranoplus cf. radamae Cataulacus MAD02 Terataner MAD02 Pheidologeton affinis Temnothorax rugatulus Metapone madagascarica Rhopalomastix rothneyi 100 Acanthoponera minor Heteroponera panamensis 90 Gnamptogenys striatula 96 100 Typhlomyrmex rogenhoferi Ectatomma opaciventre Rhytidoponera chalybaea 0.1

Figure S4 : Maximum likelihood (majority rule) inferred from the masked, partitioned data set with 5,000 bootstrap replicates (-f a; GTR + GAMMA, see method section). The tree was rooted with Pristocera. Pristocera MAD01 1 Metapolybia cingulata Mischocyttarus flavitarsis 0.94 Aglyptacros cf 0.85 Aporus niger suldatus 0.93 0.8 Dasymutilla aureola

0.99 Sapyga pumila Chyphotes mellipes Scolia verticalis outgroup taxa 1 Apis mellifera Chalybion californicum 1 1 Leptanilla GRE01 1 Leptanillinae Leptanilla RSA01 Protanilla JAP01 Martialis heureka Martialinae

0.99 1 Adetomyrma MAD02 1 Amblyopone mutica 0.78 Mystrium mysticum Amblyopone pallipes 1 Anochetus madagascarensis 0.66 Odontomachus coquereli 1 0.83 Leptogenys diminuta 1 Odontoponera transversa Pachycondyla sikorae 0.97 Centromyrmex sellaris 1 1 Loboponera politula 1 1 Plectroctena ugandensis 0.98 1 Psalidomyrmex procerus 1 Hypoponera opacior

1 Hypoponera sakalava 1 Simopelta cf. pergandei

0.69 Thaumatomyrmex atrox poneroids 1 Platythyrea mocquerysi 1 0.86 Platythyrea punctata 0.66 Paraponera clavata 1 Concoctio concenta 1 Prionopelta MAD01 Onychomyrmex hedleyi Apomyrma stygia Tatuidris ECU01

1 Discothyrea MAD07 0.87 Probolomyrmex tani 1 Proceratium MAD08 Proceratium stictum 1 Acanthostichus kirbyi 1 Cylindromyrmex striatus 1 0.73 Leptanilloides mckennae Leptanilloides nomada 0.67 0.93 Cerapachys augustae 0.99 Cerapachys larvatus 0.97 Sphinctomyrmex steinheili Simopone marleyi 1 Cerapachys sexspinus 1 1 Aenictogiton ZAM02 1 Dorylus helvolus 1 Dorylus laevigatus 1 Aenictus ceylonicus 0.99 Aenictus eugenii 0.69 1 Cheliomyrmex cf. morosus 1 Eciton vagans Neivamyrmex nigrescens 1 Myrcidris epichari 1 Pseudomyrmex gracilis 1 Tetraponera rufonigra Tetraponera punctulata Aneuretus simoni

0.99 Anonychomyrma gilberti 1 Philidris cordatus 0.95 Turneria bidentata 0.73 Azteca ovaticeps 0.61 1 Linepithema humile 0.61 1 Papyrius nitidus 0.98 0.97 Dorymyrmex bicolor Forelius pruinosus 1 Leptomyrmex AUS01 1 Leptomyrmex erythrocephalus 1 Liometopum apiculatum 1 Liometopum occidentale 1 1 Tapinoma sessile 1 Technomyrmex MAD05 Technomyrmex difficilis Dolichoderus scabridus 1 Myrmecia pyriformis Nothomyrmecia macrops 0.61 Acropyga acutiventris 0.98 0.98 Anoplolepis gracilipes Myrmoteras iriodum

1 Calomyrmex albertisi 0.83 Camponotus hyatti formicoids 1 0.97 Camponotus maritimus 1 Polyrhachis Cyrto01 1 Polyrhachis Hagio01 1 1 Camponotus BCA01 Camponotus conithorax 0.8 Opisthopsis respiciens 0.65 1 Brachymyrmex depilis 0.69 Myrmelachista JTL01 1 Oecophylla smaragdina 1 Formica moki Polyergus breviceps Notoncus capitatus 1 Notostigma carazzii 1 Lasius californicus Myrmecocystus flaviceps 0.78 Paratrechina hystrix 1 Pseudolasius australis Prenolepis albimaculata 0.76 Prenolepis imparis 0.95 1 Acanthoponera minor

1 Heteroponera panamensis 1 Ectatomma opaciventre 1 Rhytidoponera chalybaea 1 Gnamptogenys striatula Typhlomyrmex rogenhoferi

1 Manica bradleyi 0.93 1 0.99 Myrmica striolagaster Myrmica tahoensis Pogonomyrmex subdentatus 0.55 Monomorium ergatogyna 1 Myrmicaria exigua 1 1 Solenopsis molesta Solenopsis xyloni 1 Aphaenogaster albisetosa Messor andrei 0.72 1 Aphaenogaster occidentalis 1 1 Messor denticornis 0.99 Aphaenogaster swammerdami Stenamma dyscheres 0.96 Cardiocondyla mauritanica Leptothorax muscorum complex 0.92 0.98 Mayriella ebbei 0.86 0.87 Vollenhovia emeryi 0.79 Temnothorax rugatulus 0.65 Wasmannia auropunctata 1 Cataulacus MAD02 0.91 Meranoplus cf. radamae 0.56 Metapone madagascarica Nesomyrmex echinatinodis 0.92 Pheidologeton affinis

0.83 Terataner MAD02 0.99 Crematogaster emeryana Myrmecina graminicola 0.64 Eutetramorium mocquerysi 0.51 Xenomyrmex floridanus Rhopalomastix rothneyi 0.58 1 Tetramorium caespitum Tetraponera punctulata

0.91 Basiceros manni Daceton armigerum 1 Eurhopalothrix bolaui 0.53 Microdaceton tibialis Myrmicocrypta cf. infuscata 1 Pheidole clydei 0.95 Pheidole hyatti 0.9 Pilotrochus besmerus 0.64 Procryptocerus scabriusculus 1 Pyramica hoplites Strumigenys dicomas 1 Acromyrmex versicolor 0.7 Trachymyrmex arizonensis 0.64 Apterostigma auriculatum 0.98 Acanthognathus ocellatus Orectognathus versicolor

0.1

Figure S5 : Bayesian tree (majority rule consensus) inferred from the unmasked, unpartitioned data set (GTR + GAMMA, 28,130,500 generations, sample frequency 100, burn-in (10%) discarded; see method section). The tree was rooted with Pristocera. Pristocera MAD01 1 Mischocyttarus flavitarsis Metapolybia cingulata 0.99 Aporus niger 0.99 Chyphotes mellipes 1 Sapyga pumila Dasymutilla aureola outgroup taxa 1 0.88 Scolia verticalis 1 Chalybion californicum Apis mellifera 0.78 Aglyptacros cf.sulcatus

1 Protanilla JAP01 1 Leptanilla GRE01 Leptanillinae 0.98 Leptanilla RSA01 Martialis heureka Martialinae 1 Onychomyrmex hedleyi 1 1 Concoctio concenta Prionopelta MAD01 Tatuidris ECU01 Apomyrma stygia 0.82 1 0.98 Amblyopone pallipes 1 1 Amblyopone mutica Mystrium mysticum Adetomyrma MAD02 Paraponera clavata 0.52 Discothyrea MAD07 1 Probolomyrmex tani 1 1 Proceratium MAD08 Proceratium stictum 1 Platythyrea punctata Platythyrea mocquerysi Anochetus madagascarensis 'poneroids' 1 1 Leptogenys diminuta Pachycondyla sikorae 0.85 1 Odontoponera transversa 1 Odontomachus coquereli 1 Thaumatomyrmex atrox Simopelta cf. pergandei 0.94 1 Centromyrmex sellaris 0.99 Psalidomyrmex procerus 0.52 1 1 Loboponera politula Plectroctena ugandensis 1 Hypoponera sakalava Hypoponera opacior Cerapachys sexspinus 1 Aenictus ceylonicus

0.98 0.99 Aenictus eugenii

1 Cheliomyrmex cf. morosus 1 Neivamyrmex nigrescens 1 Eciton vagans 1 Dorylus laevigatus 0.81 Dorylus helvolus 0.97 1 Aenictogiton ZAM02 0.96 Cerapachys augustae

0.58 Cerapachys larvatus Simopone marleyi Sphinctomyrmex steinheili 1 Acanthostichus kirbyi 1 Cylindromyrmex striatus 1 Leptanilloides mckennae Leptanilloides nomada

1 Tetraponera punctulata 1 Tetraponera rufonigra 1 Myrcidris epicharis 0.61 Pseudomyrmex gracilis Aneuretus simoni Dolichoderus scabridus 0.87 1 Liometopum occidentale 0.9 Liometopum apiculatum 1 1 Tapinoma sessile 1 Technomyrmex difficilis Technomyrmex MAD05 1 0.64 1 Leptomyrmex AUS01 Leptomyrmex erythrocephalus 1 0.98 Dorymyrmex bicolor Forelius pruinosus

0.52 Azteca ovaticeps 0.74 Papyrius nitidus 0.73 Linepithema humile 0.74 Anonychomyrma gilberti

0.99 Turneria bidentata 0.94 Philidris cordatus 1 Myrmecia pyriformis Nothomyrmecia macrops 1 Lasius californicus Myrmecocystus flaviceps 1 Brachymyrmex depilis 0.53 Myrmelachista JTL01 formicoids Oecophylla smaragdina

1 Notostigma carazzii 1 Polyergus breviceps Formica moki 0.89 1 Polyrhachis Cyrto01

0.91 Polyrhachis Hagio01

1 Calomyrmex albertisi Camponotus hyatti 0.99 1 0.95 Camponotus maritimus 1 Opisthopsis respiciens 0.81 1 Camponotus conithorax Camponotus BCA01 Anoplolepis gracilipes Myrmoteras iriodum Notoncus capitatus Acropyga acutiventris Pseudolasius australis 1 Paratrechina hystrix Prenolepis imparis 0.86 Prenolepis albimaculata 1 Myrmica striolagaster 1 Myrmica tahoensis 0.76 Manica bradleyi Pogonomyrmex subdentatus

0.9 Daceton armigerum Orectognathus versicolor

0.79 0.99 Microdaceton tibialis 0.52 Acanthognathus ocellatus 0.72 Myrmicocrypta cf. infuscata Apterostigma auriculatum 1 Acromyrmex versicolor 0.52 0.65 Trachymyrmex arizonensis 1 Eurhopalothrix bolaui 0.55 Wasmannia auropunctata Procryptocerus scabriusculus 0.73 0.53 1 Pyramica hoplites Strumigenys dicomas 1 Pheidole clydei 0.91 Pheidole hyatti Pilotrochus besmerus Basiceros manni 1 Aphaenogaster albisetosa

0.78 Messor andrei 1 Stenamma dyscheres

0.95 1 Aphaenogaster swammerdami 1 Aphaenogaster occidentalis Messor denticornis 0.62 Metapone madagascarica 1 Solenopsis molesta 1 Solenopsis xyloni 0.99 Myrmicaria exigua 0.66 Monomorium ergatogyna

0.95 Rhopalomastix rothneyi Vollenhovia emeryi Temnothorax rugatulus 0.99 0.72 Cardiocondyla mauritanica 0.85 Leptothorax muscorum complex 1 Tetramorium validiusculum Tetramorium caespitum 0.58 0.55 0.76 Mayriella ebbei Xenomyrmex floridanus

0.88 Eutetramorium mocquerysi 0.68 Myrmecina graminicola Crematogaster emeryana Pheidologeton affinis

0.58 Terataner MAD02 Nesomyrmex echinatinodis

0.6 Meranoplus cf. radamae 1 Cataulacus MAD02 1 Acanthoponera minor Heteroponera panamensis 1 0.99 Gnamptogenys striatula 1 Typhlomyrmex rogenhoferi 1 Ectatomma opaciventre Rhytidoponera chalybaea

0.1 Figure S6 : Bayesian tree (majority rule consensus) inferred from the masked, unpartitioned data set (GTR + GAMMA, 30 million generations, sample frequency 200, burn-in (10%) discarded; see method section). The tree was rooted with Pristocera. Pristocera MAD01 1 Mischocyttarus flavitarsis Metapolybia cingulata 0.62 Aglyptacros cf. sulcatus Aporus niger 0.62 Chyphotes mellipes 0.98 Sapyga pumila outgroup taxa Dasymutilla aureola

0.83 Scolia verticalis 1 Chalybion californicum

1 Apis mellifera 1 Protanilla JAP01 1 Leptanilla GRE01 Leptanillinae Leptanilla RSA01 Martialis heureka Martialinae 1 Onychomyrmex hedleyi 1 1 Concoctio concenta

0.77 Prionopelta MAD01 Tatuidris ECU01 Apomyrma stygia 0.79 0.98 Amblyopone pallipes 1 0.98 1 Amblyopone mutica Mystrium mysticum Adetomyrma MAD02 Paraponera clavata 0.67 0.71 Discothyrea MAD07 1 Probolomyrmex tani 1 Proceratium MAD08 Proceratium stictum 1 1 Platythyrea punctata Platythyrea mocquerysi Anochetus madagascarensis 1 'poneroids' 1 Leptogenys diminuta

0.58 Pachycondyla sikorae 0.87 1 Odontoponera transversa 1 Odontomachus coquereli 1 Thaumatomyrmex atrox Simopelta cf pergandei 0.57 1 Centromyrmex sellaris Psalidomyrmex procerus

1 0.74 1 Loboponera politula Plectroctena ugandensis 0.84 1 Hypoponera sakalava Hypoponera opacior Cerapachys sexspinus Cerapachys augustae 0.74 Cerapachys larvatus Simopone marleyi 1 Dorylus laevigatus 0.96 Dorylus helvolus Aenictogiton ZAM02 1 Aenictus ceylonicus 1 0.98 Aenictus eugenii

1 Cheliomyrmex cf. morosus 1 Neivamyrmex nigrescens Eciton vagans 1 Acanthostichus kirbyi 1 Cylindromyrmex striatus 0.89 1 Leptanilloides mckennae Leptanilloides nomada Sphinctomyrmex steinheili

1 Tetraponera punctulata 0.99 Tetraponera rufonigra 1 Myrcidris epicharis

0.75 Pseudomyrmex gracilis Aneuretus simoni 1 Liometopum occidentale 0.96 0.91 Liometopum apiculatum 1 Tapinoma sessile 1 Technomyrmex difficilis 1 Technomyrmex MAD05 Dolichoderus scabridus 1 Leptomyrmex AUS01 0.51 Leptomyrmex erythrocephalus 1 0.98 Dorymyrmex bicolor Forelius pruinosus

0.63 0.73 Azteca ovaticeps Papyrius nitidus 0.73 Linepithema humile 0.74 Anonychomyrma gilberti

0.99 Turneria bidentata 0.84 0.94 Philidris cordatus 1 1 Myrmecia pyriformis Nothomyrmecia macrops 1 Lasius californicus Myrmecocystus flaviceps 1 Brachymyrmex depilis 0.6 Myrmelachista JTL01 formicoids Oecophylla smaragdina Notostigma carazzii 1 Polyergus breviceps

1 Formica moki 1 Polyrhachis Cyrto01

0.79 Polyrhachis Hagio01

0.71 1 Calomyrmex albertisi Camponotus hyatti 1 0.91 Camponotus maritimus 1 Camponotus BCA01 1 0.85 Camponotus conithorax

0.98 Opisthopsis respiciens Anoplolepis gracilipes 0.51 Myrmoteras iriodum Notoncus capitatus Acropyga acutiventris

1 Pseudolasius australis Paratrechina hystrix 0.67 Prenolepis imparis 0.9 Prenolepis albimaculata 1 Myrmica striolagaster 1 Myrmica tahoensis 0.9 Manica bradleyi Pogonomyrmex subdentatus Orectognathus versicolor 0.85 Daceton armigerum 0.9 Microdaceton tibialis Acanthognathus ocellatus Eurhopalothrix bolaui Myrmicocrypta cf. infuscata Apterostigma auriculatum 0.52 1 Acromyrmex versicolor Trachymyrmex arizonensis Procryptocerus scabriusculus 1 Pyramica hoplites Strumigenys dicomas 1 Pheidole clydei 0.94 Pheidole hyatti 1 0.52 Pilotrochus besmerus Basiceros manni 1 Aphaenogaster albisetosa

0.93 0.99 Messor andrei Stenamma dyscheres

0.98 1 Aphaenogaster swammerdami 0.95 1 Aphaenogaster occidentalis Messor denticornis Metapone madagascarica 1 Solenopsis molesta 0.9 1 Solenopsis xyloni 0.99 Myrmicaria exigua Monomorium ergatogyna Vollenhovia emeryi

0.99 Temnothorax rugatulus 0.91 Cardiocondyla mauritanica 0.5 0.88 Leptothorax muscorum complex 1 0.95 1 Tetramorium validiusculum

0.74 Tetramorium caespitum 0.59 Mayriella ebbei Rhopalomastix rothneyi Xenomyrmex floridanus 0.85 Eutetramorium mocquerysi 0.61 0.54 Myrmecina graminicola Crematogaster emeryana Pheidologeton affinis

0.77 Terataner MAD02 Nesomyrmex echinatinodis

0.59 1 Cataulacus MAD02 Meranoplus cf. radamae Wasmannia auropunctata 1 Acanthoponera minor Heteroponera panamensis 1 1 Gnamptogenys striatula 1 Typhlomyrmex rogenhoferi 1 Ectatomma opaciventre Rhytidoponera chalybaea

0.1 Figure S7 : Bayesian tree (majority rule consensus) inferred from the masked, partitioned data set (GTR + GAMMA, 30 million geneRations, sample frequency 200, burn-in (10%) discarded; see method section). The tree was rooted with Pristocera.