Taxonomic revision of Stigmatomma Roger (Hymenoptera: Formicidae) in the Malagasy region

Flavia A Esteves; Brian L Fisher

doi:10.3897/BDJ.4.e8032

. 2016 Jun 13;(4):e8032. doi: 10.3897/BDJ.4.e8032

Taxonomic revision of Stigmatomma Roger (Hymenoptera: Formicidae) in the Malagasy region

Flavia A Esteves ^‡,^✉, Brian L Fisher ^‡

PMCID: PMC4934140 PMID: 27433124

Abstract Abstract

In this study we present the first taxonomic revision of the ant genus Stigmatomma in the Malagasy biogeographic region, redescribe the previously known S. besucheti Baroni-Urbani, and describe seven new species to science (S. bolabola sp. n., S. irayhady sp. n., S. janovitsika sp. n., S. liebe sp. n., S. roahady sp. n., S. sakalava sp. n., and S. tsyhady sp. n.). The revision is based on the worker caste, but we provide brief descriptions of gynes and males for some species. Species descriptions, diagnosis, character discussion, identiﬁcation key, and glossary are illustrated with 360 high-quality montage and SEM images. The distribution of Stigmatomma species in Madagascar are mapped and discussed within the context of the island’s biomes and ecoregions. We also discuss how some morphometric variables describe the diﬀerences among the species in the bioregion. Open science is supported by providing access to R scripts, raw measurement data, and all specimen data used. All specimens used in this study were given unique identifies, and holotypes were imaged. Specimens and images are made accessible on AntWeb.org.

Keywords: Malagasy bioregion, taxonomy, ants, Amblyoponinae , Madagascar, Seychelles

Introduction

Stigmatomma Roger (1859), the largest genus of the ant subfamily Amblyoponinae, currently contains more than 40 species among extant and fossil taxa (AntCat 2016). Its systematics has recently received attention. The genus was revived as a valid after many years as a synonym of Amblyopone (Yoshimura and Fisher 2012b), and is considered to be closely related to the Amblyoponinae genera Adetomyrma, Myopopone, Mystrium, and Xymmer (a group also known as XMMAS clade; Ward and Fisher 2016, Yoshimura and Fisher 2012b, Brady et al. 2006, Ouellette et al. 2006, Saux et al. 2004). The most recent phylogenetic hypothesis on Amblyoponinae divided the genus into two groups (Ward and Fisher 2016). Fulakora, resurrected from its synonymy under Stigmatomma and elevated to generic status, is a predominantly Neotropical lineage that contains some Australasian species. The remainder species continue to be assigned to Stigmatomma. However, the phylogeny was inconclusive regarding the monophyly of the newly delimited Stigmatomma, and its relationship with the other XMMAS lineages remains debatable. We believe that a more uniform and exhaustive taxon sampling in future phylogenetic studies may help to clarify these uncertainties.

Stigmatomma is distributed globally, but very little is known about the genus in the Malagasy region, apart from the description of S. besucheti (Baroni Urbani 1978) from the Seychelles.

Madagascar and its surrounding islands are extremely biodiverse, have a high rate of animal and plant endemism, and possess exceptional rates of habitat loss due to human activity (Goodman and Benstead 2004, Myers et al. 2000). Ninety percent of the original forest cover is estimated to have disappeared from the island since the arrival of humans around 2000 years ago (Du Puy and Moat 1996, Myers et al. 2000, Goodman and Benstead 2004). The region is likely home to 1,300 ant species, of which about 60% are undescribed (Fisher 2005, Fisher 2004); furthermore, 95% of described ant species in the region are found nowhere else in the world (Fisher 2004).

Ants play a large role in a terrestrial ecosystem (Alonso 2000, Andersen 1990, Hoffmann et al. 2000), and basic knowledge of their taxonomy and distribution may provide a baseline for all subsequent research and conservation efforts including them. Brian Fisher and members of the Malagasy Arthropod Team, based at the Madagascar Biodiversity Center in Madagascar, have conducted arthropod inventories in the Malagasy region for the last twenty years in an attempt to unravel the diversity of ants in that area. They have contributed to twenty-seven taxonomic revisions, which have added more than 300 new ant species to the Malagasy ant fauna list to date. From the material they have collected, more than 900 Stigmatomma specimens have been databased, and more than 150 images of these ants are available online on AntWeb.

This study presents the first taxonomic revision for the genus Stigmatomma in the Malagasy region and recognizes eight species, of which seven are newly described. It focuses on the worker caste, but images and a brief description of gynes and males are provided when possible.

Biology

Our understanding of the biology of the species assigned to Stigmatomma is far from comprehensive as it is based on generalizations from limited observations of a few species. One of the major culprits for our lack of observations is the cryptobiotic lifestyle of these ants, which hampers access to their colonies and studies on their behavior (Ward and Fisher 2016).

Predominantly, Stigmatomma species nests in the soil or in rotten logs of humid forest habitats (Brown 1960). Workers are usually solitary hunters (Masuko 1993, Traniello 1978), but S. reclinatum (Mayr 1879), from Indomalaya region, has been found to recruit aid to recover prey (Billen et al. 2005, Ito 1993a). Stigmatomma prey upon other arthropods, especially geophilomorph centipedes (Gotwald and Lévieux 1972, Brown 1960)—observations indicate that up to 80% of the diet of S. silvestrii Wheeler 1928 (Palearctic region) is composed of such centipedes (Masuko 1993).

Larvae feed directly on prey when positioned on this food source (Brown 1960). We are not aware of any report of trophallaxis between larvae and adults of Stigmatomma. Instead, studies indicate that female adults perform nondestructive cannibalism on their own larvae. This practice, also known as Larval Hemolymph Feeding (LHF), consists of ingestion of hemolymph dripping from punctures made by adults in the larval integument. It was described for the Nearctic S. pallipes (Haldeman 1844) and for S. silvestrii (Haskins 1928, Masuko 1986, respectively). In the latter species, queens of mature colonies seem to feed exclusively on larval hemolymph. LHF was also reported for species of other Amblyoponinae genera (e.g., Amblyopone, Myopopone, Mystrium, and Prionopelta; Ito 2010, Ito and Billen 1998, Wheeler and Wheeler 1988).

The majority of the species produces winged gynes. However, some members of the Stigmatomma reclinatum species-group does not present a morphologicaclly distinct queen, and reproduction is performed by gamergates (Ito 1993b, Ito 1991). Within these colonies, dominance is established through chemical and aggressive interactions (Ito 1993b).

Materials and methods

Species delimitation

The biological species concept guides species delimitation in this study, implying that species represent reproductively isolated entities, enclosing one or many populations connected by gene flow (Coyne and Orr 2004). We thus used morphological discontinuities as evidence for species separation preferentially maintained in sympatry among closely related forms.

Species names

In this study, new species names formed from a personal name are nouns in apposition and thus invariant. All other new species names presented are not latinized words, and thus also indeclinable (ICZN article 31.2.3, ICZN 1999).

Specimen records

Every specimen we examined bears a specimen code label (e.g., CASENT0797614, ANTWEB1008502). Each code is a registered unique identifier, which aggregates several information regarding a given specimen on AntWeb (e.g., collection record, images, identification). Specimen data can be accessed on AntWeb through the persistent URL of its specimen code: www.antweb.org/specimen/“specimen code" (e.g., www.antweb.org/specimen/CASENT0797614).

Terminology

Morphological terminology used follows Keller (2011), unless otherwise stated:

Aedeagal apodemes (Snodgrass 1935): pair of anterior apodemes of the aedeagus that are connected to the majority of the aedeagal muscles (Fig. 1 b, d).
Aedeagus (Snodgrass 1935): the intromittent organ of the male genitalia. It projects from between the anterior portion of the parameres, and contains the aedeagal apodemes and the penisvalvae (Fig. 1 a, d).
Anepisternum (as in Bolton 1994): dorsal subdivision of the mesepisternum, separated from the katepisternum by the mesepisternal sulcus (Fig. 2 a).
Arolium: adhesive pretarsal organ (Fig. 3 d).
Basal ring (Snodgrass 1957): anterior annular sclerite of the male genitalia. It is connected anteriorly to the muscles from the abdominal segment IX, and posteriorly to the muscles of the aedeagus, volsellae, and parameres (Fig. 1 a, b).
Basimere (Snodgrass 1957): anterior portion of each paramere (Fig. 1 b, c).
Basivolsella (Peck 1937): anteroventral plate of the volsella; supports the cuspis and the digitus (Fig. 1 b, c).
Calcar of strigil: protibial spur, which together with the comb of strigil forms the antennal cleaning organ, which is also known as strigil (Fig. 3 a).
Comb of strigil: comb-like structure on probasitarsus, which together with the calcar forms an antennal cleaning organ (Fig. 3 a).
Cuspis (Snodgrass 1941): lobe of the volsella continuous to the basivolsella; located between the paramere and the digitus (Fig. 1 a, b, c).
Digitus (Snodgrass 1941): movable lobe of the volsella; located between the cuspis and the aedeagus (Fig. 1 a, b, c).
Epistomal sulcus: sulcus that divides the clypeus posteriorly and laterally from the remainder of the head (Richards 1977; Fig. 4 a).
Frontal lobes: in this study, we use this term to name the trait formed by: (1) the median arch projection of the torulus (torular lobe in Keller 2011); or (2) the many degrees of fusion between the anterior dorsolateral expansion of the frontal carina (posttorular flange in Keller 2011) and the median arch projection of the torulus (Fig. 4 a).
Frontoclypeal sulcus: medial section of the epistomal sulcus running between the anterior tentorial pits (Fig. 4 a).
Gaster (as in Bolton 1990): formed by the third, fourth, fifth, sixth, and seventh abdominal segments in Stigmatomma females. When mentioned separately, the Roman numeral of the corresponding homologous true segment labels abdominal segments (i.e., III– VII; Fig. 4 b).
Genal tooth (Brown 1960): lateral cuticular projection on the anterior genal angle (Fig. 4 a).
Hypopygium: sternite of abdominal segment VII in adult females (Fig. 4 b).
Katepisternum (as in Bolton 1994): ventral subdivision of the mesepisternum, separated from the anepisternum by the mesepisternal sulcus (Fig. 3 d).
Lower and upper metapleuron (Snodgrass 1910): secondary division of the metapleuron in a dorsal wing bearing part and a ventral leg-bearing part (Fig. 5 b).
Median area of the clypeus: clypeal area between and below the frontal carinae (Fig. 4 a).
Mesepimeral lobe (as epimeral lobe in Yoshimura and Fisher 2012b): posterodorsal lobe of the mesepimeron, which covers the metathoracic spiracle (Fig. 5 b).
Mesepimeron (as in Gibson et al. 1998): posterodorsal portion of the mesopleuron, which is differentiated from the mesepisternum by the mesopleural suture (Fig. 5 b).
Mesepisternum (as in Huber and Sharkey 1993): anterior subdivision of the mesopleuron, usually comprising most of the mesopleuron (Figs 2 b, 5 b).
Mesobasitarsal sulcus: longitudinal impression situated on the anterodorsal face of the mesobasitarsus (Fig. 3 b).
Mesopleural suture (Snodgrass 1910): vertical or oblique suture dividing the mesopleuron into an anterior/ventral mesepisternum and a posterior/dorsal mesepimeron. It extends from the base of the wing process to the coxal process.
Mesopleuron (as in Huber and Sharkey 1993): lateral and ventral part of the mesothorax, which is the second and largest of the three primary subdivisions of the thorax, bearing the middle pair of legs and, when present, the forewings.
Mesoscutellar-axillar complex (as in Gibson et al. 1998): region of the mesonotum posterior the transscutal articulation; often simply referred to as the scutellum, but composed of the scutellum and axillae (Fig. 5 a).
Mesoscutum (as in Gibson et al. 1998): region of the mesonotum anterior to the transscutal articulation and scutellar-axillar complex (Fig. 5 a).
Mesosoma (as in Bolton 1990): formed by three thoracic segments plus the propodeum (abdominal segment I; Fig. 4 b).
Metabasitarsal sulcus: longitudinal impression situated on the anterior face of the metabasitarsus (Fig. 3 c).
Metapleuron (as in Gibson et al. 1998): pleuron of the metathorax (Fig. 5 b).
Microtrichia: setae-like cuticular projections.
Notaulus (pl. notauli; as in Gibson et al. 1998): paired lines or grooves on the mesoscutum that subdivide the sclerite into a median midlobe and lateral lobes (Fig. 5 a).
Parameres (Snodgrass 1957): elongated pair of lateral lobes of the male genitalia; subdivided into the anterior basimere and the posterior telomere (Fig. 1 a, b, c).
Penisvalvae (Snodgrass 1941): pair of well-developed, dorsoventrally and anteroposteriorly elongated, sclerotized lateral plates of the aedeagus (Fig. 1 b, d).
Petiolar laterotergite: paired long, narrow, strip-like area of cuticle parallel to the ventral margin of the petiolar tergite (Fig. 6 a).
Petiolar proprioceptor zone: a depression sharply delineated anteriorly and bearing numerous sensilla on the anterior-most part of the petiolar sternite (Fig. 6 a).
Petiole: abdominal segment II (Fig. 4 b).
Poststernite: posterior remaining portion of sternite not concealed by an articulation (Fig. 6 b).
Pygostyles (as in Yoshimura and Fisher 2007; referred as cerci in Gibson et al. 1998): paired sensory finger-like projections that articulate with the tenth abdominal tergite of male ants.
Scuto-scutellar suture (as in Gibson et al. 1998): groove or line that separates the axillae from the scutellum (Fig. 5 a).
Subpetiolar process (as in Ward 1994): anteroventral projection of the petiolar poststernite.
Supraclypeal area (frontal triangle in Bolton 1994): well delineated and unpaired area lying immediately posterior to the median part of the clypeus, between the frontal carinae (Fig. 4 a).
Suture and sulcus: the fusion of two sclerites forms a suture, while a depression formed by an invagination of the cuticle corresponds to a sulcus.
Telomere (Snodgrass 1957): hollow posterior portion of each paramere (Fig. 1 b, c).
Transscutal articulation (as in Gibson et al. 1998): transverse line across the mesonotum at the level of the forewings that differentiates an anterior mesoscutum and posterior scutellar-axillar complex, and that permits flexion of the mesonotum for flight.
Volsellae (Snodgrass 1957): pincer-like organ located between each paramere and aedeagus. Formed by the cuspis, digitus, and basivolsella (Fig. 1 a, b, c).

Figure 1a. — llustrated glossary of terminology; plate I: male genitalia morphology.

Figure 1b. — llustrated glossary of terminology; plate I: male genitalia morphology.

Figure 2a. — llustrated glossary of terminology; plate II: mesosoma of *Stigmatomma* worker, lateral view.

Figure 2b. — llustrated glossary of terminology; plate II: mesosoma of *Stigmatomma* worker, lateral view.

Figure 3a. — Illustrated glossary of terminology; plate III: legs morphology.

Figure 3b. — Illustrated glossary of terminology; plate III: legs morphology.

Figure 4a. — IIllustrated glossary of terminology; plate IV: worker morphology.

Figure 4b. — IIllustrated glossary of terminology; plate IV: worker morphology.

Figure 5a. — Illustrated glossary of terminology; plate V: mesosoma morphology of alates.

Figure 5b. — Illustrated glossary of terminology; plate V: mesosoma morphology of alates.

Figure 6a. — Illustrated glossary of terminology; plate VI: worker morphology

Figure 6b. — Illustrated glossary of terminology; plate VI: worker morphology

Sculpture terminology follows Harris (1979) as below. In order to describe additive sculpture, we employ a dash between terms (e.g., foveate-costate means numerous pits among longitudinal costae).

Alveolate: honeycombed, with regular, deep, angular cavities separated by thin partitions; furnished with cells or alveoli.
Areolate: divided into a number of small, irregular, nonparallel spaces.
Carinate: keeled, with one, or several, but usually few longitudinal narrow raised ridges.
Catenate: with longitudinal, connected elevations like links in a chain.
Confused: indefinite outlines.
Costate: with longitudinal raised ridges (costae); coarser than carinate.
Costulate: less prominent than costate.
Dispersed: scattered markings or small sculptures.
Foveate: pitted, with numerous, regular depressions or pits (foveae).
Foveolate: with small, deep pits; finely pitted.
Imbricate: partly overlapping, like shingles on a roof or scales on a fish.
Nodulate: with small knots or swellings.
Plicate: folded; with folds.
Puncticulate: dispersed points or punctures, with very fine, widely spaced punctures.
Smooth: devoid of any sculpturing.
Strigate: with narrow, transverse raised ridges or impressed lines.
Rugose: wrinkled.
Rugulose: minutely wrinkled.
Taeniate: with broad, longitudinal ribbon-like markings; shaped like a tapeworm.
Tuberculate: furnished with rounded, projecting lobes.

We describe setae and cuticular projections with the following terminology:

Acuminate: tapering to a point (Fig. 7 a).
Antler-like: branched, shaped like an antler (Fig. 7 b).
Blunt: not sharp, a worn-down apex (Fig. 7 c).
Conic: shaped like a cone (Fig. 7 d).
Dentiform: shaped like a tooth (Fig. 7 c, d).
Digitiform: shaped like a finger (Fig. 7 e).
Filiform: shaped like a thread; filamentous. Here used to describe setae having a regular, hair-like shape (Fig. 8 a).
Flattened-apex: flattened apically, not round (Fig. 7 a).
Glabrous: devoid of hair or cuticular projections.
Lanceolate: shaped like the head of a lance (Fig. 7 e).
Mucronate: ending abruptly in a sharp point (Fig. 7 c).
Paddle-like: shaped like a paddle (Fig. 7 f).
Scrobiculate: uniformly covered with short, oblong or trench-like hollows.
Spatular: shaped like a spatula (Fig. 7 a, c).
Spiniform: shaped like a spine (Fig. 8 a).
Squamiform: shaped like a scale (Fig. 7 e).
Stout: heavily built (Fig. 8 a).
Strap-like: shaped like a long and narrow strip (Fig. 7 f).
Tongue-like: shaped like a tongue (Fig. 8 b).
Truncated: having the apex cut off transversally; lacking the apex (Fig. 7 b).
Tubiform: shaped like a tube (Fig. 8 c).

Figure 7a. — Illustrated glossary of terminology; plate VII: setae and cuticular projections.

Figure 7b. — Illustrated glossary of terminology; plate VII: setae and cuticular projections.

Figure 8a. — Illustrated glossary of terminology; plate VIII: setae and cuticular projections.

Figure 8b. — Illustrated glossary of terminology; plate VIII: setae and cuticular projections.

The terminology used to describe pilosity inclination, in regards to cuticle surface, follows Wilson (1955).

Wing venation (Fig. 9) follows Archibald et al. (2006).

Figure 9. — Diagram of the *Stigmatomma* generalized wing venation: A, forewing; B, hindwing. Abbreviations: C, costa; Sc, subcosta; R, radius; Rs, radial sector; M, media; C, cubitus; A, anal. The position of vein's free abscissas are indicated by the letter f followed by a cardinal number. Images by Masashi Yoshimura, available at AntWeb.org (specimen CASENT0083104). llustration by F. A. Esteves.

Measurements and indices

We used indices and measurements to quantify size, and as means of comparison among Stigmatomma species. Measurements were taken on a Leica MZ APO stereomicroscope, rounded to the nearest 0.01 mm. They are expressed in mm, and presented as minimum and maximum values with holotype measurements within parentheses. Indices are rounded to the nearest integer value, and expressed as minimum and maximum values with holotype index within parentheses. The raw data are presented in Suppl. material 1.

Head length (HL): in fullface view, straight line from the anterior clypeal margin to the midpoint of a straight imaginary line connecting posterior corners of the head (Fig. 10A).
Head width (HW): maximum width of the head, including eyes when present (Fig. 10A).
Head width 2 (HW2): width of the head immediately posterior to the posterolateral margin of the clypeus (as in Taylor 1978; Fig. 10A).
Scape length (SL): length of the scape (first antennal segment), excluding its basal constriction or neck and condyle (Fig. 10A).
Mandibular length (ML): outer length of the mandible (as in Taylor 1978; Fig. 10A).
Weber’s length of mesosoma (WL): diagonal length of the mesosoma in profile, from base of anterior slope of pronotum to metapleural lobe (Fig. 10B).
Propodeal posterior width (PPW): width of posterior margin of propodeal dorsal face, in dorsal view (Fig. 10C).
Petiolar length (PtL): maximum length of petiole in dorsal view (Fig. 10C).
Petiolar width (PtW): maximum width of petiole in dorsal view (Fig. 10C).
Cephalic index (CI): HW/HL ×100.
Scape index (SI): SL/HL ×100.
Mandibular index (MI): ML/HL ×100.
Petiolar index (PtI): PtL/PtW ×100.

Figure 10. — Measurements taken from *Stigmatomma* worker caste. A: fullface view; B: lateral view of head and mesosoma; C: dorsal view of mesosoma and anterior abdominal tergites. Abbreviations: HL, head length; HW, head width; HW2, head width 2; ML, mandibular length; PPW, propodeal posterior width; PtL, petiolar length; PtW, petiolar width; SL, scape length; WL, Weber's length of mesosoma. Illustrations by F. A. Esteves.

We employed a UPGMA hierarchical cluster analysis to visualize how specimens are grouped based on the differences in their linear morphometry. We also compared the clustering result with our species hypothesis to see how well they reflect each other.

In the UPGMA analysis, specimens clustered together are morphometrically more similar than specimens grouped into different clusters (Legendre and Legendre 1998). All of the following steps were performed on the R platform (R Core Team 2015; see complete script in the Suppl. material 2).

Cluster analysis uses a dissimilarity matrix as input. As a measure of dissimilarity, we used Euclidean distances, defined as the squared differences of measurement values between each pair of specimens (function dist, method “euclidean”; stats package, R Core Team 2015):

where d is the distance between specimens x and y, J is the total number of measurements taken from each specimen, and j is a given measurement.

Data normalization is imperative for unbiased Euclidean distances. It balances the contribution of each measurement to the distance matrix, neutralizing the weight of the absolute differences of larger variables (see more about data normalization in Gelman and Hill 2007). The impact of data normalization on Euclidean distances can be seen in this example of a simple model of two specimens, A and B, and three measurements (WL, PtL, and PtW):

If the same measurement values were normalized to the log scale, the Euclidean distance between specimens A and B would be only 0.0022. Thus, before data normalization, the absolute difference between Weber’s length values dominates the equation’s result. To counteract this effect, we normalize original measurement values to the natural-log scale using function log (base package, R Core Team 2015) before calculating the dissimilarity distances between specimens.

UPGMA starts by combining the couple of most similar specimens into a group, then, it adds other specimens, or combines groups to groups, until all specimens are united by a common root. We used the method "average" as clustering strategy. It combines similar clusters together using the distance among cluster centroids as a dissimilarity measurement. For clustering, we used function hclust (stats package).

We use the cophenetic correlation coefficient to measure how well the cluster represents the distances between specimens. This provides a linear correlation coefficient between the cophenetic distances obtained from the cluster, and the original dissimilarity matrix that was used to build the cluster. The output value must be close to 1 for a high-informative cluster (Oksanen 2014, Oksanen 2015). To calculate the cophenetic correlation coefficient, we use the functions cophenetic and cor (stats package), and to plot the cluster, the function plot (graphics package, R Core Team 2015).

Cluster analysis captures the multivariate structure of a dataset, but it does not unveil the patterns of variation behind the clusters it builds. We used Principal Component Analysis (PCA) to visualize and interpret patterns of morphometric variation among specimens. As seen below, PCA offers solutions for two elements that hamper the detection of patterns underlying morphometric variation among specimens: variable correlation and multidimensionality.

Morphological measurements are biologically linked to each other as they describe traits of an organism, and therefore, they are generally correlated (Zelditch et al. 2012). PCA creates new variables, or components, to eliminate such correlations. PCA components combine linearly the original measurements, are independent from each other, and will act as new dimensions/axes in the ordination space (Zelditch et al. 2012, Marhold 2011).

In the morphometric space (i.e., space defined by the measurements), PCA draws its first component along the line that comprises the highest proportion of variation among specimens. Consecutively, it derives the remaining components to encompass the highest variation after derivation of the previous components. This process continues until the number of components equals the number of original measurements (Zelditch et al. 2012, Marhold 2011).

The newly computed components are uncorrelated (i.e., orthogonal in the space), and ideally, the first components will capture most of the variation among specimens (Zelditch et al. 2012, Marhold 2011). Hence, the first components may be used to produce a graphical representation of the dataset in a lower dimensional space. The analysis provides: (1) A matrix with eigenvectors values, which are the location of original measurements on each component axis. It indicates the contribution of each original variable to the component: the larger the absolute value, the more important the variable. (2) The proportion of the total variance encompassed by each component.

Eventually, PCA projects the position of each specimen onto the components (Zelditch et al. 2012, Marhold 2011). In other words, it fits the specimens into the ordination space. The coordinate, or location, of each specimen on a given component is the PCA score (Marhold 2011).

We executed all the steps of the Principal Component Analysis on the R platform (R Core Team 2015), as follows (see complete script in the Suppl. material 3).

First, we checked the original measurements for the presence of correlations (function cor, stats package; R Core Team 2015). PCA components are based on variable correlations, which means that high correlations between variables increase the success of the analysis (Zelditch et al. 2012, Marhold 2011).

Function prcomp (scale set to TRUE; stats package) executed PCA analysis on the original measurement matrix. It also standardized our dataset to zero mean and unit variance, which prevents dominance of variables with higher variance in the analysis (Manly 2004). Function ggscreeplot (ggbiplot package; Vu 2011) produced a screeplot of the proportional variation explained by each PCA component. Function ggbiplot (ggbiplot package) mapped species scores along the components that encompassed the majority of the variation exhibited by the dataset.

The data and R scripts underpinning the analysis presented above are deposited in the Dryad Data Repository at http://dx.doi.org/10.5061/dryad.m7340.

Images

Extended focus montage images were created with a Leica DFC 425 camera and LEICA APPLICATION SUITE software (version 3.8; Leica Microsystems, Switzerland), and are available online at AntWeb. In addition, scanning electron microscopy (SEM) was used for observations of smaller characters. We prepared specimens for SEM adapting the procedure used by Keller (2011):

Workers kept in ethanol were washed in water and gently brushed to remove dirt particles, before being placed in 90% ethanol for 20 minutes. Specimens were then: (1) point mounted in a copper conductive triangle (TED PELLA, INC.) below the median and hind right coxae, and fixed in an SEM aluminum Zeiss stub (TED PELLA, INC.) via a double-sided adhesive conducting PELCO tab (TED PELLA, INC.); and (2) left to air dry for at least 12 hours before scanning.

Point-mounted dry specimens were submerged in warm water to dissolve the mounting glue before being placed in 90% ethanol, after which the same treatment described above was applied.

Specimens mounted on stubs were coated with gold-palladium—this procedure was not applied to rare taxa (i.e., poorly represented in collections). Images were taken using a LEO/Zeiss 1450 VP SEM field emission scanning electron microscope at CASC, using the high voltage mode (HV) at a voltage of 10 kV. Images of uncoated ants were taken using the SEM at a variable pressure secondary electron mode (VPSE) with the following configuration: VP target pressure around 20Pa, spot size around between 500 and 600, VPSE collector bias at 390V, and voltage at 20kV. At least three specimens of each species were imaged when permitted by the available number of specimens.

Maps

For Madagascar, species distributions were mapped over a shaded relief of the island, overlaid by an elevation layer and the outlines of five simplified ecoregion zones of the country (Burgess et al. 2004): humid forests, subhumid forests, dry deciduous forests, succulent woodlands, and spiny thickets. Mangroves were merged with the adjacent ecoregion since they are not biologically informative for Stigmatomma. The ecoregion classification used here only reflects the original primary vegetation of Madagascar. Nowadays, more than 82% of the island's original vegetation has been modified by human activities (Du Puy and Moat 1998).

For Seychelles, species distributions were mapped over a shaded-relief of the islands, overlaid by an elevation layer.

All of the following steps, unless otherwise noted, were performed on the R platform (R Core Team 2015; see Suppl. material 4 for script, which is also deposited in the Dryad Data Repository at http://dx.doi.org/10.5061/dryad.m7340).

Obtaining elevation raster layer for Madagascar: Function getData (raster package, Hijmans 2015) acquired elevation data for Madagascar, aggregating SRTM 90 m resolution data.
Obtaining elevation raster layer for Seychelles: Although function getData worked perfectly for Madagascar, it did not return any data for Seychelles. Thus, we downloaded elevation data directly from the International Centre for Tropical Agriculture (CIAT; Jarvis et al. 2008, available at srtm.csi.cgiar.org), and used function readGDAL (rgdal package, Bivand et al. 2015) to enter the data into R. Function raster (raster package) rasterized the elevation layer for Seychelles.
Obtaining and modifying ecoregions vector layers for Madagascar: In this study, ecoregion outlines of Madagascar are based on the vector data disclosed by the Terrestrial Ecoregions of the World (Olson et al. 2001, available at the WWF website). However, the original outlines were slightly mismatching the relief of Madagascar. To solve this, we combined the original ecoregion data with data from the Remaining Primary Vegetation of Madagascar (Du Puy and Moat 1996, available at the Kew Royal Botanic Gardens website), which has more natural outlines. QUANTUM GIS 1.8.0 (QGIS Development Team 2013) provided the tools to geoprocess these layers (i.e., Clip, Union, and Difference). Function readOGR (rgdal package) read the processed ecoregions files into R.
Reading distribution points for each Stigmatomma species: Function read.csv (utils package, R Core Team 2015) loaded into R a file containing the geographic coordinates of collection points for each specimen examined in this study.
Standardizing projections of raster and vector layers: Function proj4string (raster package) retrieved the vector data projection, and used it to set the projection of the elevation data. Function CRS (rgdal package) assigned that value to an R object, which was used as a liaison between retrieving and setting projections.
Obtaining the shaded relief of Madagascar: Function terrain (raster package) computed slope and terrain from the elevation data, which were used by function hillShade (raster package) to compute the shaded-relief layer.
Plottting maps: Function plot (raster package) drew the shaded relief of Madagascar and Seychelles, and overlaid them with the raw elevation layer. For Madagascar, function plot also overlaid the resuting image with the modified ecoregion layers. Function grey (grDevices package, R Core Team 2015) provided the different levels of gray for shaded-relief and elevation layers; function alpha (scales package, Wickham 2015) modified color transparency. Function points (graphics package, R Core Team 2015) drew species distributions over the map at their specified geographic coordinates.

Note that extensive myrmecological exploration of Madagascar is ongoing; we encourage readers to consult detailed and regularly updated distribution data available on AntWeb.org, where existing and future distributions can be mapped interactively and at higher resolution then the maps presented here.

Depository acronyms

ANIC: Australian National Insect Collection, Canberra, Australia.

BMNH: The Natural History Museum, London, U.K.

CASC: California Academy of Sciences, San Francisco, California, U.S.A.

MCZC: Museum of Comparative Zoology, Harvard University, Cambridge, U.S.A.

NHMB: Naturhistorisches Museum, Basel, Switzerland.

NHMW: Naturhistorisches Museum, Vienna, Austria.

MHNG: Muséum d’Histoire Naturelle, Geneva, Switzerland.

MZSP: Museu de Zoologia da Universidade de Sao Paulo, Sao Paulo, Brazil.

USNM: National Museum of Natural History, Smithsonian Institution, Washington, D.C., U.S.A.

Taxon treatments

Stigmatomma

Roger 1859

Stigmatomma as junior synonym of Amblyopone: Emery and Forel 1879: 455; Mayr 1887: 546. Revived from synonymy: Dalla Torre 1893: 14. Subgenus of Amblyopone: Forel 1900: 55; Clark 1934: 27; Brown 1949: 87. Revived status as genus: Bingham 1903: 36; Emery 1911: 23; Creighton 1950: 31. Junior synonym of Amblyopone: Brown 1960: 155. Revived status as genus: Yoshimura and Fisher 2012b: 17. Senior synonym of Arotropus: Yoshimura and Fisher 2012b: 17.
Stigmatomma = ArotropusProvancher 1881: 205. Type-species: Arotropus binodosus (junior synonym of Typhlopone pallipes), by monotypy.
Stigmatomma Stigmatomma denticulatum Roger 1859Bingham 1903: 36. by subsequent designation

Diagnosis

Workers of Stigmatomma in the Malagasy bioregion – characters of the Amblyoponinae as described by Brown (1960) and the following characters:

Mandible elongate and linear, not as long as the head, pointed at the apex (Fig. 11). Masticatory and basal margins running parallel to each other along basoapical axis, resulting in two rows of teeth (Fig. 12). Teeth of the same pair generally basally fused.
Median portion of clypeal anterior margin anteriorly projected (generally convex; Fig. 11). Anterior clypeal margin armed with single row of dentiform setae, arising from tubercle-like cuticular projections or from the flat cuticle (Fig. 12). Pair of long setae on clypeus, generally arising from its anterior margin.
Genal teeth present or absent.
Number of antennomeres: 10–12.
Under the stereomicroscope, pilosity similar present on all antennomeres (Fig. 11).
Palpal formula: 4:3; 4:2; or 2:2 (two maxillary and two labial).
Metanotal suture well developed to absent.
Mesepisternum generally divided into anepisternum and katepisternum (Fig. 2).
Number of mesotibial spurs: 0–2.
Anterodorsal face of mesobasitarsus generally with a longitudinal sulcus (Fig. 3 b).
Number of metatibial spurs: 1–2.
Anterior face of metabasitarsus generally without a longitudinal sulcus.
Pretarsal claw simple; arolium present on pro-, meso-, and metapretarsi (Fig. 3 d).
Petiole (abdominal segment II) sessile (Fig. 13). Subpetiolar process present; fenestra present or absent on its lateral face.
Constriction, generally scrobiculate, present between pretergite and postergite of abdominal segment III.
Prora present.
Scrobiculate constriction present between presclerites and postsclerites of abdominal segment IV.
Stout spiniform setae on apex of hypopygium present or absent (Fig. 14).

Figure 11. — Holotype worker of *Stigmatomma* *liebe* **sp. n.** (CASENT0318428); dorsal face of the head. Image by F. A. Esteves; available at AntWeb.org

Figure 12a. — Mandibular teeth arrangement in two different species of *Stigmatomma* in the Malagasy bioregion.

Figure 12b. — Mandibular teeth arrangement in two different species of *Stigmatomma* in the Malagasy bioregion.

Figure 13. — *Stigmatomma* *tsyhady* **sp. n.** worker; lateral view (CASENT0121332). Image by F. A. Esteves; available at AntWeb.org

Figure 14. — Abdominal segment VII of *Stigmatomma* *sakalava* worker **sp. n.** (CASENT0022146), lateral view, featuring stout spiniform setae on the apex of its hypopygium. Image by F. A. Esteves; available at AntWeb.org

Comments on worker characters

The list of characters above forms an inclusive diagnosis of the genus, but no character can currently be pointed as unique for Stigmatomma.

1. In Stigmatomma, the total dental count (including teeth arranged in pairs) recorded for Malagasy species is 11–15, distributed from base to apex as follows: 1–3 single teeth, followed by 3–6 teeth pairs (generally fused at the base), a preapical (generally single) tooth, and an apical pointy tooth (Fig. 12). Tooth number and arrangement may be constant within some species, but not for all species we evaluated: it varies within nest series and even between left and right mandibles of the same specimen. Given that, we did not use these characters alone to isolate individual species.

The most basal tooth is enlarged in the majority of species we studied, but not in all (Fig. 12). This contradicts the opinion of Yoshimura and Fisher (2012b), which is that all Malagasy Stigmatomma species possess an enlarged basal tooth in their mandibles.

Teeth coupling generally occurs between teeth with similar dimensions (Fig. 12 a). However, in two species (Stigmatomma bolabola sp. n. and S. sakalava sp. n.), dorsal teeth increase in size towards the mandible’s apex (Fig. 15). In that case, the dorsal tooth is smaller than the ventral paired tooth, but at the mandible's apex. This also contradicts Yoshimura and Fisher (2014) and Yoshimura and Fisher (2012b), who were of the opinion that dorsal teeth are smaller than ventral teeth in the XMMAS clade genera. In their view, the genus Amblyopone would generally present mandibles with no teeth pairs, but if teeth were present, the dorsal tooth would be larger than the respective ventral pair. A species noteworthy in this discussion is Stigmatomma pluto (Gotwald and Lévieux 1972) (ANTWEB1008502; Afrotropical region), whose mandible has no basal teeth paired with mandibular teeth, thus resembling the mandible of Amblyopone (Fig. 16).

Figure 15. — Mandibles of *Stigmatomma* *sakalava* **sp. n.** worker; dorsal view (CASENT0022146). Arrows point to dorsal tooth couples, which increase in size towards the mandible's apex. Image by F. A. Esteves; available at AntWeb.org

Figure 16a. — Mandibular dentition arrangement of *Stigmatomma* *pluto* worker (ANTWEB1008502).

Figure 16b. — Mandibular dentition arrangement of *Stigmatomma* *pluto* worker (ANTWEB1008502).

Among the other Amblyoponinae genera distributed in the Malagasy bioregion: Prionopelta has short and subtriangular mandibles, which are usually armed with three teeth on the apical half, so that basal and mastigatory margins are distinct (Fig. 17 c). The mandibles of Mystrium are similar to those of Stigmatomma in their indistinct basal and mastigatory margins, but are longer than its head, and have blunt apex (Bolton 1994; Fig. 17 b). Also in Mystrium, the ventral row of teeth is set far apart from the dorsal row (Yoshimura and Fisher 2014). Adetomyrma and Xymmer, like Stigmatomma, present mandibles that shorter than the head, with indistinct basal and masticatory margins and a pointy apex (Fig. 17 a, d). While teeth are not disposed in pairs along the mandibles of Adetomyrma (Yoshimura and Fisher 2012b), the mandibles of Xymmer do have pairs of teeth.

Figure 17a. — Similarities and differences of the head among *Adetomyrma*, *Mystrium*, *Prionopelta*, and *Xymmer* in the Malagasy bioregion.

Figure 17b. — Similarities and differences of the head among *Adetomyrma*, *Mystrium*, *Prionopelta*, and *Xymmer* in the Malagasy bioregion.

In addition to the similarities and differences among the shape and configuration of the mandibles, an enlarged mandibular basal tooth is absent in all other Malagasy Amblyoponinae genera (Yoshimura and Fisher 2012a, Yoshimura and Fisher 2012b; Fig. 17).

2. Number and configuration of clypeal cuticular processes and associated dentiform setae vary among the evaluated species of Stigmatomma. All species present three to ten cuticular processes on the anterior margin of the clypeus. Each medial process bears one dentiform seta.

In half of the species (tsyhady species-complex members and Stigmatomma janovitsika sp. n.), the seta on the lateral-most process is laterodistally followed by a row of dentiform setae. These lateral rows extend laterad on the anterior clypeal margin, where it arises from flat cuticle (Fig. 12 a). In few species (S. bolabola sp. n. and S. sakalava sp. n.), the lateral-most cuticular process is smaller, and does not bear any dentiform setae (Fig. 15). S. besucheti presents three medial cuticular processes that are followed laterodistally by a notch on the anterior clypeal margin. This notch is succeeded by a row of dentiform setae arising from flat cuticle (or from reduced cuticular processes; Fig. 12 b).

However, the number of medial cuticular processes may vary within some species and sometimes within nest series. Thus, we did not use such variations to isolate individual species.

Among the other Amblyoponinae genera present in the Malagasy region, Mystrium, like Stigmatomma, also presents a single row of cuticular projections bearing dentiform setae on the anterior clypeal margin (Fig. 17 b; or see ANTWEB1008554 for high-magnification images). On the other hand, Xymmer has neither specialized setae nor cuticular tubercle-like projections (Fig. 17 d; or see ANTWEB1008499 for more images); in Adetomyrma, all dentiform clypeal setae arise from flat cuticle (Fig. 17 a; or see ANTWEB1008494 for SEM images); and Prionopelta seems to have cuticular projections welded onto an anterior clypeal apron (Fig. 17 c).

A pair of long setae is present on the anterior margin of the clypeus of all genera in the XMMAS clade in the Malagasy region, however, they are reduced and stouter in Mystrium (CASENT0002095).

3. The presence or absence of genal teeth is uniform within Stigmatomma species, and this character has relative importance to group species with similar morphology. In the Malagasy bioregion, this trait is present in all Mystrium species (Fig. 17 b) and absent in Adetomyrma (Fig. 17 a), Prionopelta (Fig. 17 c), and Xymmer (Fig. 17 d).

4. Despite the variation among species, the number of antennomeres is constant within the Stigmatomma species we studied. Adetomyrma, Mystrium, and Xymmer species present no variation for this character, with all having twelve antennomeres.

5. Under the stereomicroscope, the whole antenna is equally covered by setae in Adetomyrma, Prionopelta, Stigmatomma, and Xymmer (Figs 11, 17 a, c, d). In Mystrium, the four apical-most antennomeres are covered with denser pilosity (Fig. 17 b). SEM images show that the apical antennomeres in Mystrium are actually covered by a different type of setae (ANTWEB1008554).

6. Without dissection, the maxillary and labial palpomeres are often extremely difficult to count in the species we studied.

Regarding the number of maxillary and labial palpomeres in other Amblyoponinae members in the Malagasy region, the palpal formula is constant within Mystrium (4:3) and Prionopelta (2:2) (Yoshimura and Fisher 2014), but not in Adetomyrma and Xymmer.

The palpal formula published for the Adetomyrma worker caste is 3:3, but some species are only known by the male caste, which, depending on the species, may present palpomere counts of 2:2 and 2:3 (Yoshimura and Fisher 2012a). The palpal formula for Xymmer males is 4:3/3:3/3:2 (Yoshimura and Fisher 2012b). Since published records indicate that the number of palpomeres is generally constant across castes of Amblyponinae species (Brown 1960), we expect the females of Xymmer and Adetomyrma to reflect a similarly diverse combination.

Finally, Yoshimura and Fisher (2012b) presented 4:3/4:2/3:3 as palpal formula for Stigmatomma males in the Malagasy region, differing from the numbers we counted for workers. However, mouthpart dissections on several male specimens of the same morphotypes used by Yoshimura & Fisher revealed that, for Stigmatomma, the number of palpomeres is the same in males and females (4:3/4:2; not evaluated for S. besucheti, as males are unknown).

7. The presence or absence of the metanotal suture, and the degree of its impression, may vary within species, as well as within nest series of Stigmatomma in the Malagasy region. Given this, we did not use those variations to isolate individual species.

9. The number of mesotibial spur(s) is difficult to determine under stereomicroscopes when the anterior spur is reduced in size, and also because the posterior spur may be “replaced” by an enlarged, stout spiniform seta. SEM images allowed comparisons between the texture of enlarged spinifom processes and surrounding cuticle, thus enabling us to differentiate spur and seta (Fig. 18).

Figure 18a. — Difference between a spur and an enlarged seta on the inner face of the mesotibial apex of *Stigmatomma* workers.

Figure 18b. — Difference between a spur and an enlarged seta on the inner face of the mesotibial apex of *Stigmatomma* workers.

In the Stigmatomma we studied, the number of mesotibial spurs ranged from zero to two, and were generally constant within species. In one species, S. liebe sp. n., the number of mesotibial spurs visible under the stereomicroscope ranges from one to two. The anterior spur may be visible and developed, but it is vestigial in the majority of the specimens we evaluated. This variation was observed in specimens from the same nest series.

Regarding other members of the XMMAS clade, Stigmatomma pallipes (ANTWEB1008501; Nearctic region), S. pluto (ANTWEB1008502), Adetomyrma caputleae Yoshimura and Fisher 2012a (ANTWEB1008494; Malagasy region), Fulakora mystriops (Brown 1960) (ANTWEB1008500; Neotropical region), Myopopone castanea (Smith 1860) (ANTWEB1008551; Indomalaya and Australasia regions), and Xymmer muticus Santschi 1914 (ANTWEB1008499; Afrotropical region) have two mesotibial spurs. A. venatrix Ward 1994 (Malagasy region) possesses one spur (Ward 1994), as well as F. chilensis (Mayr 1887) (ANTWEB1008496; Neotropical region) and Mystrium voeltzkowi Forel 1897 (ANTWEB1008554; Malagasy region). All Xymmer morphospecies from Madagascar evaluated under a stereomicroscope presented one spur/stout seta on the apex of the mesotibia. One species clearly seems to have a spur, while the others apparently present an enlarged stout seta.

Among the Amblyoponinae genera outside the XMMAS clade, Amblyopone australis Erichson 1842 (ANTWEB1008497; Australasia region), A. mercovichi Brown 1960 (ANTWEB1008498; Australasia region), and Apomyrma stygia Brown et al. 1971 (ANTWEB1008505) possess two mesotibial spurs. Onychomyrmex doddi Wheeler 1916 (ANTWEB1008560; Australasia region) possesses two vestigial spurs at the apex of the mesotibia. Each spur is a small, stout, conic seta totally or partially concealed by a fovea. Prionopelta aethiopica Arnold 1949 (ANTWEB1008580; Afrotropical region) and P. antillana Forel 1909 (ANTWEB1008581; Neotropical region) have one vestigial spur, while P. concenta (Brown 1974) (ANTWEB1008513; Afrotropical region) presents no spurs on the mesotibia.

10. We confirm the presence of a longitudinal sulcus on the anterodorsal face of the mesobasitarsus in all species of Stigmatomma in the Malagasy region save S. tsyhady sp. n.

Within the XMMAS clade, this sulcus is present on the mesobasitarsus of Stigmatomma pallipes (ANTWEB1008501), S. pluto (ANTWEB1008502), Adetomyrma caputleae (ANTWEB1008494), Fulakora chilensis (ANTWEB1008496), F. mystriops (ANTWEB1008500), Myopopone castanea (ANTWEB1008551), and Xymmer muticus (ANTWEB1008499). We confirm present of this sulcus in only one Xymmer species in the Malagasy region. However, this character is difficult to visualize under a stereomicroscope when specimens are too small, as it occurs with Xymmer species, and its presence or absence may be better evaluated with an SEM microscope. This sulcus is absent in all Mystrium species we evaluated in the Malagasy region (CASENT0429914; CASENT0482698; CASENT0003281; CASENT0429897; CASENT0129838; CASENT0418314; CASENT0318933; CASENT0494274; CASENT0248701; CASENT0001158; ANTWEB1008554).

The sulcus on the anterior face of the mesobasitarsus is absent in Amblyopone australis (ANTWEB1008497), A. mercovichi (ANTWEB1008498), Apomyrma stygia (ANTWEB1008505), Onychomyrmex doddi (ANTWEB1008560), Prionopelta aethiopica (ANTWEB1008580), P. antillana (ANTWEB1008581), and P. concenta (ANTWEB1008513).

11. All Stigmatomma species present in the Malagasy bioregion present two well-developed metatibial spurs save S. liebe sp. n. In this species, the number of metatibial spurs visible under the stereomicroscope ranges from one to two. The anterior spur is visibly smaller than the posterior spur, and may be vestigial in some specimens. This variation was observed in specimens from the same nest series. A similar condition is found in Onychomyrmex hedleyi Emery 1895 (Australasia region). In this species, metatibial spurs are vestigial and may be present or absent in specimens from the same colony (Brown 1960).

In the XMMAS clade, Stigmatomma pallipes (ANTWEB1008501), S. pluto (ANTWEB1008502), Adetomyrma caputleae (ANTWEB1008494), A. venatrix, Fulakora chilensis (ANTWEB1008496), F. mystriops (ANTWEB1008500), Myopopone castanea (ANTWEB1008551), Mystrium voeltzkowi (ANTWEB1008554), and Xymmer muticus (ANTWEB1008499) possess two spurs on the metatibia.

Amblyopone australis (ANTWEB1008497), A. mercovichi (ANTWEB1008498), and Apomyrma stygia (ANTWEB1008505) possess two metatibial spurs. Onychomyrmex doddi (ANTWEB1008560) possesses two vestigial spurs at the apex of the metatibia. These spurs are small, stout, conic seta totally or partially concealed by a fovea. Prionopelta aethiopica (ANTWEB1008580) and P. antillana (ANTWEB1008581) have one spur, while P. concenta (ANTWEB1008513) presents no spurs.

12. Only one Stigmatomma species evaluated in this study (S. roahady sp. n.) presents a longitudinal sulcus on the anterior face of the metabasitarsus. The metabasitarsus of S. besucheti, while not presenting a sulcus on its anterior face, possesses two raised, parallel, not-well-developed longitudinal carinae with convergent apexes on its dorsal face.

This sulcus is present on the metabasitarsus of Myopopone castanea (ANTWEB1008551), and absent in Stigmatomma pallipes (ANTWEB1008501), S. pluto (ANTWEB1008502), Adetomyrma caputleae (ANTWEB1008494), A. venatrix, Fulakora chilensis (ANTWEB1008496), F. mystriops (ANTWEB1008500), and Xymmer muticus (ANTWEB1008499). It seems to be absent on the metabasitarsus of Xymmer in the Malagasy region. However, we cautiously affirm that, since this character is difficult to visualize under a stereomicroscope when specimens are too small, like those of Xymmer, it would be better evaluated under higher magnification. This sulcus is absent in all Mystrium species we evaluated in the Malagasy bioregion (CASENT0429914; CASENT0482698; CASENT0003281; CASENT0429897; CASENT0129838; CASENT0418314; CASENT0318933; CASENT0494274; CASENT0248701; CASENT0001158; ANTWEB1008554).

Among the Amblyoponinae genera that are not part of the XMMAS clade, the sulcus on the metabasitarsus is absent on Amblyopone australis (ANTWEB1008497), A. mercovichi (ANTWEB1008498), Apomyrma stygia (ANTWEB1008505), Onychomyrmex doddi (ANTWEB1008560), Prionopelta aethiopica (ANTWEB1008580), P. antillana (ANTWEB1008581), and P. concenta (ANTWEB1008513).

13. Arolium present on pro-, meso-, and metapretarsi in all species we studied. The same seems to apply to the other Amblyoponinae genera in the Malagasy region.

14. The petiole is sessile to subsessile in Adetomyrma and Mystrium; and subsessile to penduculate in Xymmer (Fig. 19). Also, within the XMMAS clade in the Malagasy region, Xymmer is the only genus in which the subpetiolar process is absent.

Figure 19a. — Presence or absence of petiolar penduncle in *Adetomyrma*, *Mystrium*, and *Xymmer*.

Figure 19b. — Presence or absence of petiolar penduncle in *Adetomyrma*, *Mystrium*, and *Xymmer*.

15. The constriction between pretergite and postergite of the abdominal segment III is scrobiculate in all Stigmatomma species but one. In Adetomyrma such a constriction is not visible; in Xymmer species the constriction is alveolate; and in Mystrium it is scrobiculate.

16. A prora is visible under a stereomicroscope in all Stigmatomma and Mystrium species in the Malagasy region; it seems to be absent in Adetomyrma and Xymmer.

17. Adetomyrma does not possess a constriction between the presclerite and postsclerite of abdominal segment IV. The constriction is scrobiculate in Mystrium and Stigmatomma, and alveolate in Xymmer.

18. Stout spiniform setae may be located on the apex of the hypopygium, surrounding the sting (Fig. 14). The number of setae varies from six to nine, when present in Stigmatomma species in the Malagasy region. This contradicts the opinion of Yoshimura and Fisher (2014), which states that the number of stout setae ranges from three to nine.

In the Malagasy region, all Mystrium species present two or four stout setae on the hypopygium (Yoshimura and Fisher 2014), while Xymmer and Adetomyrma have no such setae. Yoshimura and Fisher (2014) affirmed that two or four stout setae on the apex of the hypopygium are uniquely observed in Mystrium; however, Fulakora mystriops (ANTWEB1008500) also presents four stout setae on the hypopygium.

Stout spiniform setae are also present on the hypopygium of Stigmatomma pluto (twelve setae, ANTWEB1008502), Fulakora chilensis (eight setae, ANTWEB1008496), and in F. cleae (Lacau and Delabie 2002) and F. agostii (Lacau and Delabie 2002), which have ten setae each (both from the Neotropical region). These setae are absent in S. pallipes (ANTWEB1008501), the Neotropical F. heraldoi (Lacau and Delabie 2002), Myopopone castanea (ANTWEB1008551), Xymmer muticus (ANTWEB1008499), Amblyopone australis (ANTWEB1008497), A. mercovichi (ANTWEB1008498), Apomyrma stygia (ANTWEB1008505), Onychomyrmex doddi (ANTWEB1008560), Prionopelta aethiopica (ANTWEB1008580), and P. concenta (ANTWEB1008513).

Malagasy species-group Stigmatomma

We introduce a morphological organization system for the species diversity of Stigmatomma in the Malagasy bioregion which is based upon the definition of informal species-groups, which may contain species-complexes when necessary. Groups and complexes are named after the most abundant species, and the groups we presently define only reflect what is seen in the Malagasy fauna.

Synoptic list of Malagasy species

besucheti group

besucheti (Baroni Urbani 1978) (Seychelles; Singapore?)

tsyhady group

sakalava complex

bolabola Esteves & Fisher sp. n. (Madagascar)

janovitsika Esteves & Fisher sp. n. (Seychelles)

sakalava Esteves & Fisher sp. n. (Madagascar)

tsyhady complex

irayhady Esteves & Fisher sp. n. (Madagascar)

liebe Esteves & Fisher sp. n. (Madagascar)

roahady Esteves & Fisher sp. n. (Madagascar)

tsyhady Esteves & Fisher sp. n. (Madagascar)

besucheti species-group

Stigmatomma besucheti (Baroni Urbani 1978)

The morphology of S. besucheti isolates the species from other Stigmatomma in the Malagasy bioregion, and we place it in its own group based on the following worker characters (asterisks flag unique characters within the genus in the Malagasy bioregion):

1. * Ten antennomeres;

2. * Two maxillary palpomeres (palpal formula: 2:2);

3. * Calcar of strigil completely pectinate;

4. * Anterior face of the calcar of strigil with squamiform microtrichia basally;

5. * Posterior face of the calcar of strigil glabrous;

6. * Weakly raised longitudinal parallel carinae present on the dorsal face of metabasitarsus, with convergent apexes;

7. * Petiolar proprioceptor zone reduced to a small concavity.

tsyhady species-group

Workers with the following combination of characters (asterisks flag unique characters within the genus in the Malagasy bioregion):

1. * Twelve antennomeres;

2. * Four maxillary palpomeres (palpal formula: 4:3 or 4:2);

3. * Calcar of strigil not completely pectinate; basoventral lamella generally visible (in one species the lamella is reduced to a basal bud);

4. Anterior face of the calcar of strigil with straplike or tubiform microtrichia basally;

5. * Posterior face of the calcar of strigil with lanceolate microtrichia;

6. Absence of any longitudinal carina on the dorsal face of metabasitarsus;

7. * Petiolar proprioceptor zone a large, round concavity.

This group can be split into subgroups based on morphological similarities, here called species complexes.

sakalava species-complex

Stigmatomma bolabola Esteves & Fisher, sp. n.

Stigmatomma janovitsika Esteves & Fisher, sp. n.

Stigmatomma sakalava Esteves & Fisher, sp. n.

Workers with the following combination of characters (character numbers are sequential to the species groups for sake of clarity in the character discussion):

8. Genal teeth present or absent;

9. Two labial palpomeres (palpal formula: 4:2);

10. Antler-like microtrichia present on posterior face of posterior metatibial spur;

11. Absence of fenestra on the subpetiolar process;

12. Stout spiniform setae present on the apex of hypopygium.

tsyhady species-complex

Stigmatomma irayhady Esteves & Fisher, sp. n.

Stigmatomma liebe Esteves & Fisher, sp. n.

Stigmatomma roahady Esteves & Fisher, sp. n.

Stigmatomma tsyhady Esteves & Fisher, sp. n.

Workers with the following combination of characters (character numbers are sequential to the species groups for sake of clarity in the character discussion; asterisks flag unique characters within the genus in the Malagasy bioregion):

8. Genal teeth present;

9. * Three labial palpomeres (palpal formula: 4:3);

10. Posterior face of posterior metatibial spur mostly glabrous;

11. * Fenestra present on the subpetiolar process;

12. * Absence of stout spiniform setae on hypopygium.

Comments on species-groups and species-complexes characters

3. A reduced lamella on the basoventral margin of the calcar of strigil is often difficult to visualize under a stereomicroscope, and the calcar may appear completely pectinate while in reality it has a basal bud on the base of its ventral margin. Nonetheless, the proportion of lamellar tissue on the ventral margin of the calcar is constant within species, and was helpful to delimit certain species.

4-5. Presence and shape of microtrichia on anterior and posterior face of the calcar of strigil are not visible under a stereomicroscope. However, those characters are informative to diagnose groups of species.

6. The longitudinal parallel carinae on the dorsal face of the Stigmatomma besucheti metabasitarsus somewhat converge at their apexes; thus, the region between them appears groove-like in dorsal view. No other species of Stigmatomma in the Malagasy region presents such a character. However, one species, S. roahady, possesses a longitudinal sulcus on the anterior face of its metabasitarsus, and since its shape and location are different from the carinae on S. besucheti, we did not consider them homologous.

10. The presence and shape of microtrichia on the posterior face of the metatibial spur are not visible under a steromicroscope; however, it is informative to diagnose groups of species.

Stigmatomma besucheti

(Baroni Urbani 1978)

Amblyopone besucheti Baroni Urbani 1978: 49, figs. 15, 16. Holotype (worker, CASENT0101816): SEYCHELLES: Ile de la Digue, 28.Jan.1975, P. Schauenberg leg. Paratypes: 9 workers; same data as holotype.
Stigmatomma besucheti Combination in Stigmatomma: Yoshimura and Fisher 2012b: 19.

Materials

Type status: Holotype. Occurrence: catalogNumber: casent0101816; recordedBy: P. Schauenberg; sex: 1w; preparations: pin; associatedMedia: http://www.antweb.org/specimen/casent0101816; Taxon: scientificName: Stigmatomma besucheti; genus: Stigmatomma; Location: country: Seychelles; locality: Ile de la Digue; decimalLatitude: -4.359097; decimalLongitude: 55.841242; georeferenceRemarks: coordinates obtained from Google Earth; Event: eventDate: 01/28/1975; Record Level: institutionCode: MHNG
Type status: Paratype. Occurrence: catalogNumber: casent0101900; recordedBy: P. Schauenberg; sex: 1w; preparations: pin; associatedMedia: http://www.antweb.org/specimen/casent0101900; Taxon: scientificName: Stigmatomma besucheti; genus: Stigmatomma; Location: country: Seychelles; locality: Ile de la Digue; decimalLatitude: -4.359097; decimalLongitude: 55.841242; georeferenceRemarks: coordinates obtained from Google Earth; Event: eventDate: 01/28/1975; Record Level: institutionCode: MHNG
Type status: Paratype. Occurrence: catalogNumber: casent0101970; recordedBy: P. Schauenberg; sex: 1w; preparations: pin; associatedMedia: http://www.antweb.org/specimen/casent0101970; Taxon: scientificName: Stigmatomma besucheti; genus: Stigmatomma; Location: country: Seychelles; locality: Ile de la Digue; decimalLatitude: -4.359097; decimalLongitude: 55.841242; georeferenceRemarks: coordinates obtained from Google Earth; Event: eventDate: 01/28/1975; Record Level: institutionCode: MHNG
Type status: Paratype. Occurrence: catalogNumber: casent0280650; recordedBy: P. Schauenberg; sex: 1w; preparations: pin; otherCatalogNumbers: BMNH(E)1017525; associatedMedia: http://www.antweb.org/specimen/casent0280650; Taxon: scientificName: Stigmatomma besucheti; genus: Stigmatomma; Location: country: Seychelles; locality: Ile de la Digue; decimalLatitude: -4.359097; decimalLongitude: 55.841242; georeferenceRemarks: coordinates obtained from Google Earth; Event: eventDate: 01/28/1975; Record Level: institutionCode: BMNH
Type status: Paratype. Occurrence: catalogNumber: casent0906833; recordedBy: P. Schauenbert; sex: 1w; preparations: pin; associatedMedia: http://www.antweb.org/specimen/casent0906833; Taxon: scientificName: Stigmatomma besucheti; genus: Stigmatomma; Location: country: Seychelles; locality: Ile de la Digue; decimalLatitude: -4.359097; decimalLongitude: 55.841242; georeferenceRemarks: coordinates obtained from Google Earth; Event: eventDate: 01/28/1975; Record Level: institutionCode: NHMB
Type status: Paratype. Occurrence: catalogNumber: casent0906834; recordedBy: P. Schauenberg; sex: 1w; preparations: pin; associatedMedia: http://www.antweb.org/specimen/casent0906834; Taxon: scientificName: Stigmatomma besucheti; genus: Stigmatomma; Location: country: Seychelles; locality: Ile de la Digue; decimalLatitude: -4.359097; decimalLongitude: 55.841242; georeferenceRemarks: coordinates obtained from Google Earth; Event: eventDate: 01/28/1975; Record Level: institutionCode: NHMB
Type status: Paratype. Occurrence: catalogNumber: casent0906835; recordedBy: P. Schauenberg; sex: 1w; preparations: pin; associatedMedia: http://www.antweb.org/specimen/casent0906835; Taxon: scientificName: Stigmatomma besucheti; genus: Stigmatomma; Location: country: Seychelles; locality: Ile de la Digue; decimalLatitude: -4.359097; decimalLongitude: 55.841242; georeferenceRemarks: coordinates obtained from Google Earth; Event: eventDate: 01/28/1975; Record Level: institutionCode: NHMB
Type status: Other material. Occurrence: catalogNumber: casent0172194; recordedBy: D.H.Murphy; sex: 1w; preparations: pin; otherCatalogNumbers: anic32-016286; associatedMedia: http://www.antweb.org/specimen/CASENT0172194; Taxon: scientificName: Stigmatomma cf. besucheti; genus: Stigmatomma; Location: country: Singapore; locality: Univ. Campus Singapore; (ANIC32-016286); decimalLatitude: 1.28333; decimalLongitude: 103.767; georeferenceRemarks: 10km; Identification: identifiedBy: R.W. Taylor; dateIdentified: 12/06/1978; Event: eventDate: 06/01/1964; habitat: Imperata grassland; eventRemarks: soil; Record Level: institutionCode: ANIC
Type status: Other material. Occurrence: catalogNumber: casent0195513; recordedBy: D.H.Murphy; sex: 1dq; preparations: pin; otherCatalogNumbers: anic32-016287; associatedMedia: http://www.antweb.org/specimen/CASENT0195513; Taxon: scientificName: Stigmatomma cf. besucheti; genus: Stigmatomma; Location: country: Singapore; locality: Bukit Timah Nature Reserve; decimalLatitude: 1.38333; decimalLongitude: 103.8; georeferenceRemarks: 10km; Identification: identifiedBy: R.W. Taylor; dateIdentified: 12/07/1978; Event: eventDate: 10/04/1965; habitat: degraded coastal hill forest; eventRemarks: on grantite, berlesate No I24; Record Level: institutionCode: ANIC

Description

Worker (Fig. 20): HL: 0.38-0.40; HW: 0.28-0.29; HW2: 0.26-0.27; SL: 0.18-0.19; ML: 0.20-0.22; WL: 0.41-0.43; PPW: 0.15-0.16; PtL: 0.14-0.15; PtW: 0.16-0.16; CI: 71.36-73.54; SI: 46.04-48.68; MI: 52.26-55.45; PtI: 90.00-92.50.

Figure 20a. — Holotype of *Stigmatomma* *besucheti* (CASENT0101816); worker. Images by April Nobile; available at AntWeb.org

Figure 20b. — Holotype of *Stigmatomma* *besucheti* (CASENT0101816); worker. Images by April Nobile; available at AntWeb.org

Head:

Mandibular baso-masticatory margin skirted dorsally by row of filiform setae; medially, by flexuous tongue-like setae; ventrally, by flexuous filiform setae, grading into flexuous tongue-like setae apically (Fig. 21 a, b). Mandibular dentition arrangement, from base to apex: two single teeth (same size of teeth arranged in pairs); four pairs of teeth (each pair with same dimensions, fused basally); single preapical tooth; apical tooth (Fig. 21 a). Pair of teeth with similar dimensions along mandible's basoapical axis. Anterior clypeal margin with three cuticular processes arranged in a single row, armed anteriorly with asymmetrical mucronate dentiform seta; followed laterally by a notch on the anterior clypeal margin (Fig. 21 a). Most lateral portion of anterior clypeal margin armed with row of conical setae arising from flat cuticle (or from reduced tubercle-like cuticular processes), decreasing in size laterad. Clypeal cuticular processes with approximately same length of associated dentiform setae. Long filiform setae pair on clypeal median area, posterior to central-most pair of cuticular processes on clypeal anterior margin. Shorter filiform pair of setae on clypeal median area, between longer pair of setae and frontal lobes. Median area of clypeus extending posteriorly between antennal sockets as narrow longitudinal strip; frontoclypeal sulcus acute (Fig. 21 c). Supraclypeal area as small oblong depression (Fig. 21 c). Ten antennomeres. Genal teeth absent. Compound eyes absent. Palpal formula: 2:2 (two maxillary, two labial; Fig. 21 d).

Figure 21a. — Head of *Stigmatomma* *besucheti* worker.

Figure 21b. — Head of *Stigmatomma* *besucheti* worker.

Mesosoma:

In dorsal view, mesonotum somewhat expanded laterally (Fig. 22 a). Metanotal suture absent. Sulcus dividing mesepisternum into anepisternum and katepisternum; dorsoposterior corner of katepisternum not rounded (Fig. 22 b). Metathoracic spiracle slit-like, surrounded by cuticular swell, projected posteriorly, inserted in a concavity. Propodeal spiracle round, surrounded by a cuticular swell. Propodeal declivitous face slightly concave (Fig. 22 a).

Figure 22a. — Mesosoma of *Stigmatomma* *besucheti* worker (CASENT0101970). Images by F. A. Esteves; available at AntWeb.org.

Figure 22b. — Mesosoma of *Stigmatomma* *besucheti* worker (CASENT0101970). Images by F. A. Esteves; available at AntWeb.org.

Legs:

Absence of lamella on basoventral margin of calcar of strigil (Fig. 23 a). Calcar of strigil anterior face with squamiform microtrichia basally (Fig. 23 a); posterior face mostly glabrous (Fig. 23 b). Multiple paddle-like setae on anteroventral face of protibial apex, next to calcar of strigil. Multiple paddle-like setae on anterior face of probasitarsus; row of stout setae along posterior face, next to comb of strigil. Mesotibial spur absent; apex of mesotibial inner face with deep fovea (Fig. 23 c). Slit-like longitudinal sulcus on anterodorsal face of mesobasitarsus, with apical end projected laterally (Fig. 23 c). Two metatibial spurs; anterior spur simple with lanceolate microtrichia; posterior spur pectinate (Fig. 23 d). Anterior face of posterior metatibial spur glabrous (Fig. 23 d); posterior face with antler-like microtrichia dorsoapically (Fig. 23 e). Dorsal face of metabasitarsus with two parallel carinae with convergent apexes (Fig. 23 f). Brush of few stout paddle-like setae on baso-inner face of metabasitarsus (Fig. 23 e). Arolium on pro-, meso-, and metapretarsus.

Figure 23a. — Legs of *Stigmatomma* *besucheti* worker.

Figure 23b. — Legs of *Stigmatomma* *besucheti* worker.

Metasoma:

Petiole sessile (Fig. 24 a). Ventroanterior margin of petiolar tergite anterior dorso-latero-ventral carina (Ward 1990) 2x the size of anterior margin of subpetiolar process, in lateral view (Fig. 24 a). Subpetiolar process with obtuse angle at midpoint of its ventral margin (fin-like; Fig. 24 a). Absence of fenestra on lateral face of subpetiolar process. Petiolar proprioceptor zone reduced to small round concavity bearing few sensilla (Fig. 24 b). Prora present (Fig. 24 c). Smooth sulcus between pretergite and postergite of abdominal segment III; weakly scrobiculate sulci between presclerites and postsclerites of abdominal segment IV (Fig. 24 c). Five to six stout spiniform setae on apex of hypopygium (Fig. 24 d).

Figure 24a. — Petiole and gaster of *Stigmatomma* *besucheti* worker (CASENT0101970). Images by F. A. Esteves; available at AntWeb.org.

Figure 24b. — Petiole and gaster of *Stigmatomma* *besucheti* worker (CASENT0101970). Images by F. A. Esteves; available at AntWeb.org.

Sculpture:

Mandibular dorsal face costate-slightly catenate basally, grading into costate apically except for smooth apical portion (Fig. 21 a). Clypeal median area smooth, grading into costate to rugose laterally (Fig. 21 a). Supraclypeal area smooth (Fig. 21 c). Head, in dorsal view, taeniate-catenate; area posterior to tentorial pit plicate (Fig. 21 c). Labrum weakly imbricate (Fig. 25 a). Mesosoma in dorsal view and lateral face of pronotum foveolate (Fig. 22). Anepisternum smooth; katepisternum mostly imbricate; metapleuron costate posteriorly; lateral face of propodeum smooth, grading into imbricate posteroventrally (Fig. 22 b); declivitous face of propodeum smooth (Fig. 22 a). Petiolar tergite imbricate ventroanteriorly, grading into weakly and scarcely foveate dorsally and posteriorly (Figs 24 a, 25 b). Petiolar laterotergite imbricate; petiolar poststernite mostly areolate, grading into smooth ventrally (Fig. 24 b). Most of gaster weakly foveolate (Fig. 25 c).

Figure 25a. — Body sculpture in *Stigmatomma* *besucheti* worker (CASENT0101970). Image by F. A. Esteves, available at AntWeb.org.

Figure 25b. — Body sculpture in *Stigmatomma* *besucheti* worker (CASENT0101970). Image by F. A. Esteves, available at AntWeb.org.

Pilosity and color:

Suberect pilosity on head, dorsal face of mesosoma, lateral face of pronotum, petiolar tergite, and abdominal segments III, IV, V, and VI. Legs densely covered by subdecumbent pilosity. Suberect pilosity on anterior half of petiolar poststernite. Longer pilosity on abdominal segment VII. Body color yellow.

Comments on character variation

Character variation among specimens was minimal.

Other castes

Unknown for the Malagasy region. However, a gyne of a putative Stigmatomma besucheti is known for Singapore, as presented below.