Then there were five: a reexamination of the ant genus Paratrechina (Hymenoptera, Formicidae)

Abstract The ant genus Paratrechina is reexamined based on the discovery of two new species from Madagascar (P. ankarana sp. n. and P. antsingy sp. n.). Paratrechina kohli, a species known from central Africa, is transferred to Paratrechina from Prenolepis based on a new morphological interpretation of the genus and an updated morphological diagnosis of the genus is provided. This means that other than the widespread P. longicornis, whose origins remain uncertain, all Paratrechina are restricted either to the Afrotropical or Malagasy regions. It would also appear that of the five Paratrechina species now known, three are from dry forest habitats. With this reexamination of the genus, the possible origins of P. longicornis are discussed. A key to the genera of the Prenolepis genus-group is provided, as is a key to the workers of Paratrechina. In addition, we also designate a lectotype for Paratrechina kohli.


Introduction
During the course of revising the ant genera Paraparatrechina (LaPolla et al. 2010c) and Nylanderia (LaPolla and Fisher, in prep) in Madagascar, two species were discovered that were initially thought to belong to the genus Prenolepis. This was particularly interesting because the genus is unknown from the Malagasy region. Despite recent definition changes to Prenolepis, following the removal of the Neotropical species to their own genus, Zatania , the genus remains poorly understood taxonomically. There were two features of these two new species from Madagascar that gave us pause as to their potential placement in Prenolepis: they did not possess the characteristic mesonotal constriction immediately behind the pronotum that is present in most Prenolepis, and the eyes were towards the midline of the head, whereas most Prenolepis have high eyes relative to the midline of the head (LaPolla et al. 2010a). Given that these same characters are variable within Zatania, it was assumed Prenolepis was displaying a similar level of variability. However, to be certain, the DNA of one of the new Malagasy species was sequenced (due to sample's old age this was impossible for the second species) and compared against a large dataset of other formicine ants (Ward, unpublished data). The molecular data strongly suggested that the species from Madagascar was not a Prenolepis but rather belonged in Paratrechina. This in retrospect makes morphological sense because it renders the mesosomal constriction coupled with high eyes and long palps an unequivocal diagnostic feature of Prenolepis. It also provides for a more complete understanding of morphological variation observed within Paratrechina.
Here we describe those two species in the genus Paratrechina and make some definition changes to the genus, including moving a species from Prenolepis to Paratrechina, based on the discovery of these new species. We also provide an updated key to the genera of the Prenolepis genus-group.

Materials and methods
Specimens examined for this study are deposited in the following institutions:

CASC
California Academy of Sciences, San Francisco, USA PBZT Parc Botanique et Zoologique de Tsimbazaza, Antananarivo, Madagascar USNM National Museum of Natural History, Washington, DC, USA All measurements were taken at 80× power with a Leica M125 microscope using an orthogonal pair of micrometers, recorded to the nearest 0.001 mm, and rounded to two decimal places for presentation. Multiple specimens were measured for each species, and minimum and maximum measurements and indices are presented. All measurements are given in millimeters. Digital color images were created using a Lei-ca DFC425 digital camera. Leica Application Suite software (ver. 3.8) was used for images. Each imaged specimen is uniquely identified with a specimen-level unique identifier (e.g. CASENT0454372). Morphological terminology for measurements and indices employed throughout are defined as (following LaPolla et al. 2011a, LaPolla et al. 2011b

Results and discussion
LaPolla et al. (2013) provided a diagnosis for Paratrechina, which was based on the only two species known from the genus at that time: P. longicornis (Latreille, 1802) and P. zanjensis LaPolla, Hawkes & Fisher, 2013. Based on both morphological similarity and molecular data (Ward, unpublished data), it would appear that those two species are sister taxa, so the earlier diagnosis provided for the genus was morphologically restrictive compared to what we now know is a morphologically more diverse genus. This emphasizes that there is a need to keep documenting new species in this genus-group, in particular in some of the smaller genera (i.e. Euprenolepis, Zatania) where new species might cause some additional morphological definition changes. In more speciose genera, such as Nylanderia and Paraparatrechina, this appears to be less of a concern, as the morphological definitions provided (LaPolla et al. 2010a) have not needed to be amended despite the recent discovery of dozens of new species in each genus LaPolla et al. 2010b;LaPolla et al. 2011b). The addition of three species to Paratrechina not only allows for the more complete morphologically based definition provided, but it also simplifies the morphological diagnosis of Prenolepis. With the removal of P. kohli from Prenolepis, the mesonotal constriction immediately behind the pronotum becomes a characteristic of all Prenolepis species. Such a constriction is also seen in the Southeast Asian genus Euprenolepis (La-Polla 2009), as well as in the Caribbean and Mesoamerican genus Zatania, but only in three of the five known species .
With our placement of P. kohli into Paratrechina and the description of two endemic Malagasy species, the center of Paratrechina diversity is now squarely in the Afrotropical and Malagasy regions, which raises some interesting questions as to the origin of the now pantropical P. longicornis (LaPolla et al. 2013). A review of the argument that P. longicornis is Asian in origin is provided by Wetterer (2008) and LaPolla et al. (2013) (i.e., on the observation that P. longicornis has only been found in undisturbed habitats in tropical Asia). However, since P. zanjensis appears to be a miombo woodland specialist and its sister taxon is almost certainly P. longicornis, it raises the possibility that P. longicornis might be an African woodland specialist as well. Given the fact that most African woodland habitat has been impacted to some extent by humans, it might be difficult to prove that P. longicornis is in fact native there. It is worth noting that the two new Malagasy species described here are native to dry forest habitats on limestone outcrops. While certainly P. kohli is a rainforest species, we cannot dismiss the possibility that it is the only species native to rainforests in the genus. There is one report of P. longicornis from native forest in Cameroon (Dejean et al. 1996). However, only one specimen of P. longicornis (out of 62,708 specimens) was collected from 15 forest sites in Tanzania (P. Hawkes, pers. comm.), so the conclusions of the previous study remain equivocal. Clearly, the question of the native range of P. longicornis remains an open one, but with the discovery of multiple Paratrechina species in the Afrotropical and Malagasy regions, a more complete survey is needed, and an Asian origin for the species now seems questionable.

Diagnosis of the genus
For only one species are all castes known (P. longicornis); therefore we provide only a worker-based diagnosis for the genus.
Monomorphic, medium sized (2.1-3.2 mm in total length); ranging from almost black to brownish-yellow, with lighter mandibles, antennae (especially funicular segments towards tips) and legs (especially distal portion of tibiae and tarsi). Head with medially erect macrosetae roughly paired, extending through the medial portion of clypeus. Antennae 12 segmented; scapes long, with scape index above 140, in most species around or above 200 (SI range 143-226). Scapes with a dense layer of pubescence. Head is usually distinctly longer than wide, with cephalic index below 100 (CI range 71-94); posterolateral corners rounded, with straight posterior margin. Eyes large relative to head width (REL2 greater than 25); eyes distinctly convex, extending beyond head margin in full-face view. Mandibles in all species, except P. kohli, with 5 teeth; in P. kohli 8 teeth present, with 7 th tooth on basal angle of mandible and 8 th tooth on inner mandibular margin; mandalus large and anteriorly placed; palps very long; palp formula 6:4. Mesosoma elongated, most robust in P. kohli; most gracile in P. longicornis; propodeal dorsal face variable from either nearly flat (P. longicornis) or distinctly convex (P. antsingy); propodeum without macrosetae, anteriorly occasionally with a sparse layer of pubescence; pronotal setal count 6-12 (both sides of notum); mesonotal setal count 4-8 (both sides of notum). In lateral view, petiole cuneate, broadly rounded dorsally, with much longer posterior face and not surpassing the height of the propodeum. Legs distinctly long (profemur length 0.6-1.0 mm). Gaster robust, covered in abundant erect macrosetae.

Updated worker-based key to the genera of the Prenolepis genus-group
This key is modified from that found in LaPolla et al. (2012).  (Nylander, 1856). Synonymy with longicornis by Roger 1863: 10 = Paratrechina longicornis hagemanni (Forel, 1901  Dark brown; antennae lighter with trochanters yellow to white; cuticle smooth and shiny, except on propodeum which posseses fine striations across surface; under microscrope view cuticle has a faint greenish-blue reflection; abundant macrosetae across scapes, head, pronotum, mesonotum, and gaster; scapes with short pubescence. Head ovate; posterolateral corners rounded with complete posterior margin; midpoint of eyes at approximately midline of head; eyes convex; 3 small ocelli present; mandibles with 5 teeth; apical tooth the longest, 3 rd tooth from apical shortest, remainder about the same size; outer mandibular surface with slight striations across surface. In lateral view, pronotum rises in a straight margin towards mesonotum with only slight convexity towards mesonotal margin; propodeum large and bulbous, making division between dorsal and declivitous faces difficult; metanotal suture with distinct impression that extends down along the mesopleural/metapleural margin; mesopleuron with darkened ridge along impression.

Paratrechina ankarana
Etymology. The specific epithet is derived from the name of the reserve where the species was found.  CASENT0906916 (CASC). 12 paratype workers with same locality information as holotype (BMNH, CASC, USNM).
Worker diagnosis. Cuticle smooth and very shiny; brown with patches of lighter cuticle across body, giving species a mottled appearance; SI less than 195.
Etymology. The specific epithet is derived from the name of the province where the species was found. (Forel, 1916), comb. rev.

Figs 7-9
Prenolepis kohli Forel, 1916: 438 (worker and queen decribed Light brown with darker gaster; head and mesosomal cuticle densely rugorecticulate; head, pronotum, mesonotum and gaster with scattered macrosetae; scapes without macrosetae, with layer of dense pubscence. Head subrectangular; posterolateral corners with slight angles; most of eyes above midline of head; eyes convex; no ocelli apparent; mandibles with with 8 teeth, one tooth on basal margin, another on inner mandibular margin; outer mandibular surface with slight striations across surface. In lateral view, pronotal margin nearly straight, at less than 45° angle rise towards mesonotum; pronotal/mesonotal margin with mesonotum raised slightly above margin; propodeum convex with rounded dorsal face and longer, steep declivitous face; metanotal suture with distinct impression.

Notes on Paratrechina species
Workers of Paratrechina are relatively easy to identify to species based on morphology. In addition to the color and sculpture differences listed in the key and descriptions, the two Malagasy species have shorter scapes than P. longicornis and P. zanjensis (SI less than 195) relative to their head widths. The widespread P. longicornis has the lowest profile of any Paratrechina with an almost flat pronotum, mesonotum, and propodeum. Scape macrosetae are variable within the genus with both P. longicornis and P. kohli lacking them altogether, but P. ankarana, P. antsingy and P. zanjensis have abundant scape macrosetae. The REL2 is higher in P. longicornis and P. zanjensis than in the other species, typically well into the upper thirties and low forties. For the other three species the REL2 is typically in the upper twenties and low thirties, reflecting an overall wider head relative to eye length.
Only P. longicornis and P. zanjensis workers come closest to resembling one another and these can be easily separated based on the present or absence of scape macrosetae. Workers of all species, except P. kohli, have three small, but distinct, ocelli present. The number of mandibular teeth on P. kohli is particularly interesting. The tooth on the inner mandibular margin is separated from the basal angle tooth by a short diastema. It would not be surprising to occasionally find a P. kohli worker with more than 8 teeth because the single queen specimen known in collections has a tooth that is divided with two sets of cusps, implying more teeth might occasionally be expressed in individuals. In all other Paratrechina workers only five mandibular teeth are present. Males are known only from Paratrechina longicornis (LaPolla et al. 2013), so it is impossible at this time to discuss general male characteristics for this genus.