The following Museums and research institutions provided loans or images or otherwise assisted with making material available for this study:

AMS Australian Museum, Sydney, Australia.

ANIC Australian National Insect Collection, Canberra, ACT, Australia.

BMNH The Natural History Museum, London, UK.

CASC California Academy of Sciences, San Francisco, California, USA.

HNHM Hungarian Natural History Museum, Budapest, Hungary.

JDM Curtin Ant Collection, Curtin University of Technology, Perth, Western Australia, Australia (now amalgamated under WAM–see below).

MCZ Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts, USA.

MHNG Muséum d’Histoire Naturelle, Geneva, Switzerland.

MSNG Museo Civico di Storia Naturale ’Giacomo Doria’, Genoa, Italy.

NHMB Naturhistorisches Museum, Basle, Switzerland.

QM Queensland Museum, Brisbane, Queensland, Australia

SAMA South Australian Museum, Adelaide, South Australia, Australia.

SMNH Naturhistoriska riksmuseet, Stockholm, Sweden

WAM Western Australian Museum, Perth, Western Australia, Australia.

ZHMB Museum für Naturkunde der Humboldt Universität, Berlin, Germany.

Novel type material (holotypes and paratypes) was deposited in the following institutions: ANIC, BMNH, MCZ, QM and SAMA. Note: Although it did not provide material or images, reference is also made in this monograph to The Tropical Ecosystems Research Centre (TERC), centred in Darwin, Northern Territory, Australia.

Major and minor workers of each species have been included in the measurements where possible. However, for some uncommon species only minor or major workers are known. Size and shape characters were quantified and are reported as lengths or indices. Measurements were made with a stereomicroscope using a dual-axis stage micrometer wired to digital readouts. All measurements were recorded in thousandths of a millimetre, but expressed here to the nearest hundredth. The following measurements and indices are reported (readings are presented in mm).

EL Eye length: maximum length of the eye measured in profile

HL Head length: maximum head length in full-face (dorsal) view, measured from the anterior-most point of the clypeal margin to the midpoint of a line drawn across the posterior margin of the head.

HW Head width: maximum head width measured in full-face (dorsal) view, excluding the eyes.

ML Mesosomal length: mesosomal length measured from the anterior surface of the pronotum proper (excluding the collar) to the posterior extension of the propodeal lobes.

MTL Maximum tibial length: maximum length of the tibia of the middle leg, excluding the proximal part of the articulation which is received into the distal end of the femur.

PpH Propodeal height: maximum height of propodeum measured tangentially from a line drawn between the metanotal groove to the posterior extension of the propodeal lobes.

PpL Propodeal length: length of propodeum measured from the metanotal groove to the posterior extension of the propodeal lobes.

SL Scape length: scape (first antennal segment) length excluding the basal neck and condyle.

CI Cephalic index: head width (HW)/head length (HL) × 100.

EI Eye index: eye length (EL)/head width (HW) × 100.

SI Scape index: scape index: SL/HW × 100.

Tissue collection. Specimens were collected by hand and placed into 100% ethanol.

DNA extraction. A non-destructive DNA extraction method, ANDE (Castalanelli et al. 2010) was used to extract DNA from all the specimens used in this study.

Amplification and Sanger sequencing. For specimens M001-M154 and M303-M356, amplification of the five genes (COI, LR, Wg, AA, and H3) was performed using the primers outlined in Table 1. PCR amplification volumes were 25 µl, including 2.5 µl of ANDE extracted DNA (1:10 dilution DNA:water) which was the equivalent of 20 to 25 ng/µl. The reaction mix comprised of 1 × polymerase buffer (Bioline; MyTaq), 0.2 µM of each primer. The thermocycler conditions were: 95 °C for 10 min; 40 cycles of: 95 °C for 30 s, annealing for 30 s, and 72 °C for extension, with a single final extension.

Quality and quantity of the amplified PCR products were determined on a 1.5% agarose gel. Sequencing of the amplified genes was carried out by AGRF (Australian Genomic Research Facility) using an Applied Biosystems ABI 3730 48-capillary DNA analyser with Big Dye Terminator Technology according to the manufacturer’s protocols (Applied Biosystems).

Amplification and amplicon sequencing. For specimens M155-302, amplification of the five genes gene was performed using the primer outlined in Table 1, with the additional M13 sequences 5'-GTAAAACGACGGCCAGT-'3 and 5'-AACAGCTATGACCATG-'3, added to the 5' end of the forward and reverse sequences, respectively. PCR volume was 25 µl, including 2.5 µl of ANDE extracted DNA (1:10 dilution DNA:water) which was the equivalent of 20 to 25 ng/µl. The reaction mix comprised of 1 × polymerase buffer (Bioline; MyTaq) and 0.2 µM of each primer. The thermocycler conditions were: 95 °C for 10 min, 40 cycles of: 95 °C for 30 seconds, annealing for 30 seconds, and 72 °C for extension; with a single final extension of 72 °C for 30 s for 5 minutes.

To prepare the amplicons for next generation sequencing a second round of PCR amplification was performed. The primers consisted of the Roche Lib-A adapters (Roche), a unique multiplex identifier (MID) sequence, and the complementary M13 sequence. In total 96 unique primer sets were constructed so that batches of 96 samples could be pooled and analysed at once. PCR volume was 25 µl, including 2.5 µl of first round amplifcation extracted DNA (1:100 dilution DNA:water). The reaction mix comprised 1 × polymerase buffer (Bioline; MyTaq) and 0.2 µM of each primer. The thermocycler conditions were: 95 °C for 10 min; 10 cycles of: 95 °C for 30 seconds, annealing for 30 seconds, and 72 °C for extension; with a single final extension.

Quality and quantity of the amplified PCR products were determined on a Labchip GXII (Perkin Elmer) using the HT 1K LabChip (Perkin Elmer). For each gene, the successfully amplified PCR products were normalised to 1ng/uL and pooled. Each gene pool was purified using the E-Gel® SizeSelect™ Gels (Thermofisher), then further purified using Agencourt AMPure XP (Beckman Coulter) as per the manufacturers’ instructions. The quantity of each gene was determined on a Labchip GXII (Perkin Elmer) using the HT 1K LabChip (Perkin Elmer) and each gene mixed to an equimolar concentration, term library. The library was diluted 1 in 1,000 and further purified twice using Agencourt AMPure XP as per the manufacturers’ instructions. The library was quantified using quantitative PCR as per Bunce et al. (2012) and diluted to approximately 2 million copies. The GS Junior sequencing run was set up as per the manufacturers’ instructions, and run for 200 flows.

Data analysis. The data were exported in fasta format and analysed using a self-written bioinformatics script based in Python (Python Software Foundation). Each sequence was analysed for its MID and primer sequence, which allowed the sequence to be assigned back to its specimen identification number and gene with all associated sequences saved in a single fasta file. For each specimen, the fasta file for each gene was imported in Geneious research software (BioMatters) and assembled and checked for sequencing errors. Low-occurring single nucleotide polymorphisms (SNP) were ignored, but where an SNP was >30% an IUPAC (International Union of Pure and Applied Chemistry) code was used. Each gene was translated using the standard invertebrate mitochondrial and nuclear codes, where applicable, to detect the presence of nuclear mitochondrial DNA (NUMTs).

Phylogenetic analysis. Phylogenetic relationships among each of the complexes and species were assessed using *BEAST (Drummond et al. 2012). To test the complexes proposed here, individuals were grouped according to their complex and similar species were grouped together according to their morphological identification. Each of the five genes was systematically tested using the substitution models General Time Reversible (GTR), Hasegawa-Kishino-Yano (HKY), Jukes-Cantor (JC) and clock models (strict and lognormal). Due to the high divergences within the COI gene, this data set was split and the first and second codon bases were examined with and without the third base. Each of these combinations were evaluated with gen=10,000,000, log=1000. Evolutionarily Stable Strategy (ESS) scores were evaluated for each run to determine which model was the best for each gene. The final tree models were HKY for the genes AA, Wg, and H3, and GTR for COI and LR. Strict clock rates were run for all genes for 50,000,000 generations and logged every 5,000 generations.

BEAST (Drummond and Rambaut, 2007) was used to estimate temporal distance between target taxa and their most recent common ancestor (Time to Most Recent Common Ancestor, or TMRCA). This program is suitable for evaluating rooted, time-based phylogenies using strict or relaxed molecular clocks; it is also useful as a framework for testing evolutionary hypotheses by focusing on multiple tree topologies, rather than merely a single tree. BEAST uses the Markov Chain Monte Carlo (MCMC) algorithm sampling method to average across tree space (i.e., the dimension that includes all possible trees), so that each tree is weighted proportionally to its posterior probability. In our case, only four out of the five genes were selected due to the availability of other Formicinae sequences deposited on genbank (NCBI). The evolutionary clock model Generalised Time Reversible (GTR) was applied to the LR and COI genes. In contrast, the Hasegawa, Kishino and Kano (HKY) model was applied to the AA and Wg genes. These models were selected as they each provided optimal ESS scores. No Melophorus fossils are known (http://antwiki.org/wiki/Checklist_of_Melophorus_species), so were not available for this exercise. The first node, including the root height parameters, was offset to 145 million years ago (Ma ago) with an exponential prior distribution and a mean of 10 Ma, which returned quantiles of 145.5 and 175.0 Ma ago, i.e., 5% and 95% respectively. The lower boundary represented the date calculated by Brady et al. (2006), while the upper boundary corresponded to the MRCA boundary for ants calculated by Moreau et al. (2006). The second node calibrated the MRCA for the Melophorini group, which was set with an exponential prior distribution offset to 52.6 Ma ago (the lowest date calculated by Blaimer et al. (2015) and mean 7 Ma, which returned an upper quantile of 78.42 Ma ago. This upper bound coincides with the vicariance event of Australia splitting from New Zealand and New Caledonia (80–60 Ma ago) (Stevens 1989, Yan and Kroenke 1993). These bounds are also supported by the placement of the South America genus Lasiophanes within Melophorini (Ward et al. 2016).

Forty-six taxa from the M. aeneovirens, M. biroi, M. ludius and M. potteri species-groups produced useable sequences (Table 2). No representatives of the M. anderseni or M. majeri species-groups were available for analysis, and a single representative of the M. fulvihirtus species-group failed to produce a useable sequence The number of samples amplified for each gene ranged from 116 (AA) to 141 (Wg). In total, 83 individuals from 33 species were successfully sequenced for all five genes, and these constitute most of the species discussed in the section dealing with molecular analysis.

Within the genus Melophorus the COI gene fragment was the most divergent with a maximum interspecific divergence of 17.3% (Table 3.). Similarly, the intraspecific divergence was four time higher than that for the nuclear genes. Several species had COI intraspecific divergences >5%; these included M. bagoti, M. biroi, M. perthensis, M. rufoniger, M. sulla, M. turneri, and M. wheeleri. As expected, the nuclear genes were considerably more conserved than the mitochondrial gene and ranged from an average intraspecific distance of 0.87% to 1.89% for AA and H3, respectively. The minimal divergences within the nuclear genes are reflected in the lack of clarity they bring when included in the species trees.

Overall, the results of the species-level tree (Figure 4) are concordant with the morphology. The species tree reveals well supported clades at the complex level with the most prominent exception being M. chauliodon (which aligns with the M. fieldi complex based on the genetics, although the mandibular morphology of the major worker would seem to place it within the M. wheeleri complex). However, a three-gene species tree (Suppl. material 1) constructed using the LR, Wg and COI genes strongly places M. chauliodon within the M. wheeleri complex. In fact, our conclusions based on the morphological appearance of the taxa is that the three-gene tree appears to corroborate some of the relationships suggested by the morphology more than does the five-gene tree. This may be due to confusion arising from uninformative signals from the histone and Abdominal-A genes in the latter. Thus, in the three-gene tree, M. fulvidus and M. chauliodon (as mentioned above) are placed near to taxa with which they share anatomically significant structures. In the five-gene tree they appear out of place. Indeed, most species anatomically associated with the M. wheeleri complex and the M. fieldi complex also show corresponding associations within the three-gene tree. The placement of M. xouthos and M. parvimolaris in the M. wheeleri complex, however, is supported by the five-gene tree but not the three-gene tree.

A significant exception to the general concordance of morphology and genetics is the position of the taxa Melophorus hirsutus, M. pusillus and M. translucens on both three-gene and five-gene species trees. The clade including these three ants exhibits an unexpectedly basal relationship to indisputable members of the M. biroi species-group. This is counterintuitive, as their morphology would support a close relationship between these taxa and those within the M. mjobergi-biroi clade, let alone the broader species-group. However tempting it might be to suggest that sequencing issues were responsible for this unexpected result, the algorithm used produced the same disconcerting result for individual gene trees as well as multiple gene trees, so that it cannot readily be thought of as artificial (but see further discussion under Evolutionary history discussion).

The sister-species relationships are in contrast to the generally strongly supported broader evolutionary patterns; for example, the relationships within the M. fieldi and M. wheeleri complexes are only weakly supported. The lack of support reflects the choice of genes: the mitochondrial gene was strongly divergent with high levels of homoplasy at the third codon base. When analysis was conducted with the third base incorporated, several species became polyphyletic and the ESS scores dropped below 200. However, the removal of the third base dramatically increased the ESS scores and rendered almost all species monophyletic, apart from M. biroi, M. turneri, M. sulla, and M. hirsutipes.

Given that the removal of the third codon base still left consistent paraphyly or even polyphyly for M. turneri, M. sulla and M. hirsutipes, other explanations for this phenomenon are required. The most obvious one is incorrect identification. (We confess here that the inaccurate identification of these sometimes maddeningly similar little ants accounted for many early sequencing discrepancies before good diagnostic characters could be discovered!) However, even after careful checking and rechecking, the three taxa mentioned continued to pose problems requiring a different explanation. Firstly, the splitting could be indicative of sibling or cryptic species not identified by morphological analysis. The high pairwise divergence among the three species (> 5%) might suggest this explanation. This is less likely to be the case with M. hirsutipes in which the degree of genetic divergence is less pronounced (7.8% compared with 11.8% in M. turneri and 8.9% in M. sulla), although there is a greater degree of physical variability among M. hirsutipes workers than there is in the other two taxa. In the case of M. hirsutipes also, the three-gene tree showed results (e.g., the placement on the tree of samples M153 and M323) that were disparate compared with the results of other trees. In the three-gene tree (strongly influenced by the COI configuration), M153 and M323 are on different branches, but they are sisters in the Wg tree. Similar results occurred with M. turneri. This suggests possible hybridisation. The senior author found likely hybrids of M. turneri and M. bruneus north of Wiluna, and incomplete speciation or, at least, relatively high gene flow between more-or-less discrete species, may also account for the polyphyly seen. Thirdly, since genotype does not automatically match phenotype, the trees could be illustrating incomplete lineage assortment, with more than one allele found in the genes used for a given species represented in the sequencing. Since these three species are all very common ants that occur across a wide range of ecosystems throughout Australia, incomplete lineage assortment is a possibility that should also be investigated.

Possibly one or a combination of these three interpretations (and there are other possibilities) could account for the differences seen on the trees for these three similar but morphologically discrete species. Perhaps the greatest likelihood, given their strongly derived position in the three-gene tree, is that the incidence of multiple alleles through frequent mutations in the mitochondrial gene COI due to rapid evolution is very high in these taxa. The field evidence of possible hybridization in this Melophorus complex may also indicate the fixation of the speciation process is not yet complete for M. hirsutipes, M. sulla and M. turneri (this could also be the explanation for separation of samples for the same taxon in the slower evolving Wg tree).

The issues surrounding M. biroi (which occupies a much more basal position on the five-gene tree) are more substantial, and this taxon is likely a complex of two or more species (displayed in Figure 5). This matter is discussed further under the individual species accounts.

The phylogenetic history of Formicidae and, to a lesser extent, Formicinae, has been discussed extensively by Moreau et al. (2006) and Brady et al. (2006). These two groups of authors suggest that TMRCA of formicoids/poneroids and leptanillines lived between 116 (Brady et al. 2006) and 156 (Moreau et al. 2006) Ma ago, with the advent of the Formicinae ranging from 77 (Brady et al. [2006]) to 92 (Moreau et al. [2006]) Ma ago. More recently, Blaimer et al. (2015) has provided further insight into the Formicinae and its constituent tribes. Our chief aim in this work was to estimate the divergence dates for the species-groups within Melophorus (Table 4).

Based on the estimated ages of this study, we propose that the Formicinae arose significantly earlier than the 77 Ma ago estimated by Brady et al. (2006), and closer to the dates estimated by Moreau et al. (2006) and Blaimer et al. (2015). Furthermore, formicine fossil records suggest a MRCA exceeding 92 Ma ago (Grimaldi and Agosti 2000), which is within the parameters of our data. The MRCA of 57.03 Ma ago for Melophorini (congruent with Blaimer et al. [2015]) coincides with the vicariance events that led to the separation of Australia from New Zealand and New Caledonia (80-60 Ma ago [Stevens 1989, Yan and Kroenke 1993]). The inclusion of the genus Lasiophanes by Ward et al. (2016), which is endemic to South America, also supports this date. For organisms other than ants, the link between South America and Australia flora and fauna and New Zealand has been well documented for this period (e.g., mostly extinct lineages of southern beech (Nothofagus) (Cook and Crisp 2005), marsupials (Mitchell et. al 2014, Nilsson et al. 2004, Nilsson et al. 2010), and Diptera (Yeates and Wiegmann 2005). The other Melophorini are almost entirely restricted to Australia. The sole exceptions are Prolasius advena, which is endemic to New Zealand but has Australian affinities (http://www.landcareresearch.co.nz/publications/factsheets/Factsheets/prolasius-advenus), and Notoncus gilberti, which is widespread on the mesic east and south-west coasts of Australia. This species has also been recorded from New Guinea (http://www.antwiki.org/wiki/Notoncus_gilberti) but could conceivably have been introduced by human agency. Within the Melophorini the origins of Melophorus were estimated to be ~35Ma ago, again congruent with Blaimer et al. (2015).

Within the Melophorus genus, M. hirsutus shares a MRCA with the remainder of the species sampled that is estimated at ~35 Ma ago. Not only does this ant have several unique attributes (i.e., the virtual absence of a metanotal groove, a translucent lamina surrounding the pronotum, and distinctive conical or sub-conical eyes and protrusive antennal sclerite [this latter seen only in one population of another, unrelated Melophorus species]), it is phylogenetically basal and remains one of the few Melophorus to favour a mesic environment. Within Australia, this species is confined to the wet east coast and is not found much west of the Great Dividing Range. During the period in which Melophorus hirsutus appears to have arisen, Australia was experiencing moist climatic conditions but started to dry out during the middle of the Miocene (23-5 Ma ago). Temperatures became more elevated in the period 21-14 Ma ago. After 14 Ma ago, temperatures dropped sharply. Acidification of waterways in central Australia continued to increase and river systems reduced their output and became salt lakes (White 1994).

Based on our molecular findings, this coincides with the period in which rapid diversification within the M. fieldi complex occurred (Figure 6). The M. fieldi complex is the most speciose group within Melophorus. These findings corroborate other studies that support the hypothesis that the aridification of Australia has led to rapid diversification of plants, marsupials, and various reptile groups (see table 1 in Byrne et al. 2008).

aeneovirens species-group

aeneovirens complex

aeneovirens (Lowne 1865)

=constans Santschi 1928syn n.

=insularis Wheeler 1934syn n.

=iridescens (Emery 1887) syn n.

=iridescensfraudatrix Forel 1915syn n.

=iridescensfroggatti Forel 1902syn n.

attenuipes sp. n.

canus sp. n.

castaneus sp. n.

clypeatus sp. n.

curtus Forel 1902

fulgidus sp. n.

gibbosus sp. n.

griseus sp. n.

kuklos sp. n.

mullewaensis sp. n.

platyceps sp. n.

praesens sp. n.

rufoniger sp. n.

sulconotus sp. n.

tenuis sp. n.

teretinotus sp. n.

bagoti complex

bagoti Lubbock 1883

=Camponotus cowlei Froggatt 1896comb. rev.

gracilipes sp. n.

nemophilus complex

nemophilus sp. n.

anderseni species-group

anderseni Agosti 1998

andersenioides sp. n.

chrysus sp. n.

subulipalpus sp. n.

biroi species-group

biroi complex

argus sp. n.

biroi Forel 1907

=fieldipropinqua Viehmeyer 1925syn. n.

=marius Forel 1910syn. n.

castanopus sp. n.

compactus sp. n.

cuneatus sp. n.

dicyrtos sp. n.

graciliceps sp. n.

gracilis sp. n.

latinotus sp. n.

lissotriches sp. n.

longiceps sp. n.

macrops sp. n.

microreticulatus sp. n.

minimus sp. n.

mjobergi Forel 1915

postlei sp. n.

propebiroi sp. n.

turbineus sp. n.

brevignathus complex

brevignathus sp. n.

marmar sp. n.

quadratus sp. n.

fieldi complex

ankylochaetes sp. n.

bruneus McAreavey 1949

eumorphus sp. n.

fieldi Forel 1910

fulvidus sp. n.

gilliatensis sp. n.

hirsutipes sp. n.

incisus sp. n.

inconspicuus sp. n.

isaiah sp. n.

lanuginosus sp. n.

longipes sp. n.

major Forel 1915 stat. n.

microtriches sp. n.

orthonotus sp. n.

paramorphomenus sp. n.

perthensis Wheeler 1934 stat. n.

sericothrix sp. n.

setosus sp. n.

solitudinis sp. n.

sulla Forel 1910 stat. n.

turneri Forel 1910

=pillipes Santschi 1919syn n.

=turneriaesopus Forel 1910syn. n.

=turnericandidus Santschi 1919syn. n.

vitreus sp. n.

oblongiceps complex

oblongiceps sp. n.

wheeleri complex

brevipalpus sp. n.

caeruleoviolaceus sp. n.

cerasinoniger sp. n.

chauliodon sp. n.

diversus sp. n.

hexidens sp. n.

laticeps Wheeler 1915

parvimolaris sp. n.

pelorocephalus sp. n.

prominens sp. n.

purpureus sp. n.

(‘species K’ [TERC])

wheeleri Forel 1910

=omniparens Forel 1915syn. n.

xouthos sp. n.

fulvihirtus species-group

barbellulatus sp. n.

fulvihirtus Clark 1941

ludius species-group

hirsutus complex

hirsutus Forel 1902

ludius complex

ludius Forel 1902

pusillus sp. n.

translucens sp. n.

majeri species-group

majeri Agosti 1998

potteri species-group

macroschismus sp. n.

pelecygnathus sp. n.

potteri McAreavey 1947

Total: 93 species

Removed from Melophorus

scipioForel 1915 (to Prolasius) comb. rev.

The key to species is based on workers, as species-level separation of the other adult castes is not only extremely difficult, but of very limited utility. Moreover, the reproductive castes of the less common species have either not been collected or are often unable to be associated with workers (since most material examined has come from trapping or opportunistic hand collection, rather than from nests). Major and minor workers of a specific Melophorus taxon have been included in the measurements where possible. However, for some uncommon species only minor or major workers are known. (Note. ‘species K’ (TERC) and other Melophorus specimens held in the TERC Collection were not available as loan material except under restrictive conditions that were not acceptable to the authors of this paper.

Descriptions. (Unless otherwise stated, specimens mentioned under ‘other material examined’ are housed in the Australian National Insect Collection [ANIC]. Curtin Ant Collection material [designated ‘JDM’] is now the property of the Western Australian Museum [WAM]).

This includes what are probably the most morphologically generalised Melophorus, with ancient traits. The M. aeneovirens species-group is comprised of three complexes (aenovirens, bagoti, and nemophilus). Most members of the group are elongate and gracile and include the predominant Melophorus species in cooler, wetter and more heavily forested environments, especially on the east coast, as well as parts of the Torresian zone. While some taxa have broad geographic ranges, others are known from a few collections only and appear to be highly localised. With the exception of a cluster of species around M. curtus (i.e., M. praesens, M. rufoniger and M. tenuis) the species are readily recognisable and do not pose great taxonomic difficulties.

Phylogenetically, the genetics of the M. aeneovirens species-group are highly congruent with the morphology. Despite only successfully sequencing several species, we obtained a representative species for each of the species complexes and here show that each complex is monophyletic. The M. aeneovirens complex is the most basal complex (well supported by the three- and five-gene trees), and the M. nemophilus and M. bagoti clades are sister complexes, which is reflected in the similarity in their morphological features.

Among all the M. aeneovirens complexes, this is the most speciose. The members of the complex are grouped together, based on several morphological features (particularly, the appearance of the anterior margin of the clypeus, the mandible, placement of the ammochaetae and the sloping appearance of the propodeum). The three-gene tree (Suppl. material 1) is the most telling when it comes to explaining the phylogeny of this complex. Based on the information derived from the successfully sequenced species, M. canus and M. griseus are sister species and the most highly derived. Interestingly, M. praesens is sister to M. platyceps in this gene tree, despite significant morphological differences (flattened body, short labial palps, etc.) that appear to place M. platyceps in its own complex.