Exploring the diversity of the Malagasy Ponera (Hymenoptera: Formicidae) fauna via integrative taxonomy

The genus Ponera includes over 60 extant species worldwide. These tiny, endogeic predator ants are predominantly distributed in the Indomalaya and Australasia regions, with a few additional Holarctic species. Herein, we explore and describe the diversity of the Malagasy Ponera fauna through an integrative taxonomic approach. We obtained our morphological species hypotheses from multivariate analyses of ten continuous morphometric characters. Species boundaries and reliability of morphological clusters were tested via confirmatory Linear Discriminant Analysis (LDA), cross-validation (LOOCV), and analyses of a mitochondrial COI gene fragment. According to the combined application of the analyses, altogether, three species are inferred in the Malagasy region, Ponera petila Wilson (1957), P. swezeyi Wheeler (1933), and P. adumbrans Csősz & Fisher sp. n. Ponera petila and P. swezeyi belong to the Indo-Australian Ponera tenuis group; the third species, P. adumbrans sp. n., is morphologically similar to the Papua New Guinean P. clavicornis Emery (1900). Furthermore, Linear Discriminant Analysis classified the type specimens of P. bableti Perrault (1993), along with a P. petila cluster with posterior p = 1. Therefore, we propose the new junior synonymy of P. bableti with P. petila. Madagascar’s extant biodiversity is predominantly explained by colonization events from the African continent across the Mozambique channel via rafting. However, since no native Ponera species are known from the Afrotropical continent, and the closest congeners have an almost exclusively Indo-Australian distribution, the likelihood of an Indo-Australian origin of the Malagasy Ponera fauna is implied.


Introduction
The Malagasy region, i.e., Madagascar and the surrounding smaller islands, harbors unique fauna.Madagascar's fauna is taxonomically imbalanced (Samonds et al., 2013).Many lineages of its extant terrestrial and freshwater assemblages can be characterized by high levels of species endemism (Goodman & Benstead, 2005).Within some other clades, Madagascar's fauna is species-poor.This is true for ants as well, with the exception of some hyperdiverse genera, e.g., Tetramorium (Hita-Garcia & Fisher, 2014), Camponotus (Rakotonirina et al., 2016;Rasoamanana et al., 2017), and Pheidole (Salata & Fisher, 2021).Other genera, such as the genus Ponera, are represented by only a few species in the region.The genus Ponera, encompassing tiny, endogeic predator ants, is primarily distributed in the Indomalaya and Australasia regions, with a few additional Holarctic species (Leong et al., 2019;Taylor, 1967;Wilson, 1957) and currently contains 60 extant and five fossil species worldwide (https:// www.antcat.org/).The taxonomy of the handful of Ponera species in the Malagasy region has never constituted the basis of focused research before.
We, for the first time, revise the Malagasy Ponera species via an integrated taxonomic protocol incorporating morphometrics, morphology, and DNA sequence data.Quantitative analyses of continuous morphometric traits were performed via NC-PART clustering as described by Csősz and Fisher (2016a, b).The clusters returned by this protocol are considered species unless this hypothesis conflicts with other biological information.Species boundaries and reliability of morphological clusters were tested via confirmatory Linear Discriminant Analysis (LDA), cross-validation (LOOCV), subjective evaluation of qualitative morphological characteristics, and analyses of a mitochondrial COI gene fragment.According to the combined application of the analyses, altogether, three species are inferred, Ponera petila Wilson (1957), P. swezeyi Wheeler (1933), and P. adumbrans Csősz & Fisher sp. n.Two of the three known Malagasy Ponera species, P. petila and P. swezeyi, belong to the Indo-Australian Ponera tenuis group; the third species, P. adumbrans sp.n., is morphologically similar to the Papua New Guinean P. clavicornis (Emery, 1900), that further strengthens the Indo-Australian connection hypothesis.Beyond Ponera, five other Malagasy ant genera, Adelomyrmex, Aphaenogaster, Chrysapace, Eurhopalothrix, and Vollenhovia, exhibit this biogeographic pattern, i.e., are not present in Africa but are distributed in India or the Indo-Australian regions (Fisher & Bolton, 2016).
Madagascar's extant biodiversity is predominantly explained by colonization events from the African continent across the Mozambique channel via rafting once lower sea levels allowed for the journey in the Cenozoic era (Ali & Huber, 2010;Godfrey et al., 2020;Masters et al., 2020).However, the occurrence of Malagasy Ponera species raises issues about the colonization route of the Malagasy Ponera, because (i) there are no known native extant or extinct Ponera species known from the Afrotropical continent, and (ii) both the tenuis group and clavicornis group, to which the Malagasy Ponera belong, almost exclusively exhibit an Indo-Australian distribution.Therefore, an Indo-Australian origin of the Malagasy Ponera fauna should be considered.In the present study, 10 continuous morphometric traits were recorded in 154 worker individuals belonging to 90 nest samples from the Malagasy, Indo-Australian, and Pacific regions.

Ant
The material is deposited in the following institutions, abbreviations after Evenhuis (2013): CASC (California Academy of Sciences, San Francisco, California, USA), MCZC (Museum of Comparative Zoology, Cambridge, Massachusetts, USA), MNHN (Muséum National d'Histoire Naturelle), PSWC (Phil S. Ward's collection, University of California Davis, Davis, California, USA), BPBM (Bernice P. Bishop Museum, Honolulu, Hawaii, USA), HNHM (Hungarian Natural History Museum).Type material and samples that were morphometrically investigated are presented in the "Type material investigated" and "Material examined" sections in the format as follows: CASENT code, collection code, verbatim locality, longitude, latitude, elevation in meters, collector, date (number of individuals measured, abbreviation of depository).Additional information on habitat and microhabitat, whenever available, is given in brackets.For the full list of material, see Supplementary Table S1.All images of specimens are available online on AntWeb (https:// www.antweb.org/).Images are linked to their specimens via the unique specimen identifier affixed to each pin CASENT0159966.Digital color montage images were created using a JVC KY-F75 digital camera and Syncroscopy Auto-Montage software (version 5.0), or a Leica DFC 425 camera in combination with the Leica Application Suite software (version 3.8).In verbatim descriptions of the Malagasy Ponera species, we follow terminology and character states established by Bolton and Fisher (2011).

Morphometric character recording
All measurements were taken by the first author using a Leica M165C stereomicroscope equipped with an ocular micrometer at a magnification of × 120.Body size dimensions are expressed in µm.Due to the abundance of worker specimens relative to queen and male specimens, the present revision is based on the worker caste only, which can further be justified by the fact that the name-bearing type specimens in this study belong to the worker caste.For the definition of morphometric characters, earlier protocols (Csősz & Seifert, 2003) were considered.Explanations and abbreviations for measured characters are given in Table 1.Morphometric data for each measured specimen are given in Supplementary Table S2.
Further indices that have consistently been applied by earlier researchers (Perrault, 1993;Taylor, 1967;Wilson, 1957) are also provided in this work in order to improve integration of the new findings to earlier species concepts: PNI Petiolar node index (PEW/MW × 100)

Statistical framework-hypothesis formation and testing
Generating prior species hypotheses via the combined application of NC clustering and PART.This method searches for discontinuities in continuous morphometric data, sorts all similar cases into the same cluster (see Seifert et al., 2014), and determines the ideal number of clusters.The present statistical framework follows the procedure applied in Csősz and Fisher (2016a, b).Advantages and limitations of the present procedure are discussed there.
Arriving at final species hypothesis using confirmatory LDA and LDA ratio extractor.To provide increased reliability of species delimitation, hypotheses for clusters and classification of cases via exploratory processes were confirmed by LDA leave-one-out cross-validation (LOOCV).Analyses were done in R (R Core Team, 2020).

DNA sampling
We sequenced 658 base pairs (bp) of the mitochondrial cytochrome oxidase I (COI) gene from 182 Malagasy Ponera specimens.DNA extraction and COI sequencing were performed at the University of Guelph (Ontario, Canada) following the protocol described in Fisher and Smith (2008).An additional 60 Ponera sequences from different biogeographic regions, including Australasia and Indomalaya, and 3 outgroup sequences (Diacamma, Parvoponera, and Pseudoponera) were obtained from GenBank.Outgroup selection was based on Ponera phylogeny from Branstetter and Longino (2019).GenBank accession code and additional information for all sequences are given in Supplementary Table S3.

Molecular phylogenetic inference
We used molecular phylogenetic inference to confirm species taxonomic identification.Sequences were aligned using the Geneious Alignment algorithm implemented in Geneious 6.1.8(Biomatters Ltd.).We included only unique haplotypes from each locality, resulting in a final alignment with 86 sequences.Phylogenetic analyses were performed using Bayesian inference in Mr. Bayes 3.2.7aand Maximum Likelihood using RAxML-NG online platform (https:// raxmlng.vital-it.ch/).The best-fit model of molecular substitution was estimated using PartitionFinder2 (Lanfear et al., 2016) under the Bayesian Information Criteria and linked branch lengths.The best partitioning scheme selected has 3 subsets, partitioning by codon position, as follows: GTR + G for first position, F81 for second position, and TRN + G for third position.Maximum likelihood bootstrap consensus tree was estimated using SumTrees 4.0.0 (Sukumaran & Holder, 2010, 2015).For Bayesian inference, two independent analyses of Metropolis coupling MCMC algorithm (MC 3 ) ran for 10 6 generations, sampling every 500 generations, resulting in 1500 trees after a burn-in of 25%.Each analysis ran with 4 chains, 1 cold and 3 heated at a temperature of 0.1 and chain swap frequency of 1 each generation.Convergence of independent runs was assumed when the standard deviation of split frequencies reached values below 0.01.

Species concept
Integration of the evidence provided by different lines of evidence into an integrative species hypothesis is performed based on the principles advocated by Schlick-Steiner et al.
(2010) and "integration by congruence" protocol introduced by Padial et al. (2010).In addition, we employ the universal GAGE species concept formulated by Seifert (2020), which represents the most widely used and accepted theoretical and practical indications in ant taxonomy.

Results
The exploratory NC-clustering, in combination with two hypothesis-free partitioning methods "hclust" and "kmeans," concordantly returned three completely separated clusters (Fig. 1) in Malagasy Ponera samples based on 10 morphometric characters.The confirmatory LDA including the pessimistic LOOCV confirmed this pattern showing complete classification success.The type specimens (four syntype specimens of P. swezeyi are nested in one of the clusters, while the holotype of P. petila and the type series of P. bableti are nested together in another).These types have also been classified along with the relevant cluster with posterior p = 1, when type specimens were set to wildcard in LDA, i.e., no grouping label was added for type specimens.The third cluster endemic to Seychelles is considered a new species.Though this species clearly belongs to the P. clavicornis group, no similar available species could be identified using the literature (Leong et al., 2019;Perrault, 1993;Taylor, 1967;Wilson, 1957).This species is named Ponera adumbrans sp.n. reflecting its restriction to Seychelles islands.Phylogenetic analyses corroborate this finding, showing three independent and monophyletic clusters from the Malagasy region (Fig. 2).We found complete agreement between monophyly of clusters in the molecular phylogeny and the quantitative morphology-based cluster delimitation protocols.Taking all of these lines of evidence into account, we consider the three clusters to be three species.
The three Ponera species distributed in the Malagasy region also differ in qualitative diagnostic features and body ratios (for body ratios, see Table 2).

Etymology
The specific name "adumbrans" is a Latin singular present active participle in the nominative case that means "obscuring or silhouetting."It refers to the obscure origin of this species in the Malagasy region and the type locality, Silhouette Island.long, with a rather short and weakly convex dorsum.Dorsal region of petiole smooth to finely punctate, shiny.Anterior and posterior faces of node usually clearly convergent dorsally.Lateral surface of node meets the posterior surface in a rounded transition without a cuticular ridge or transverse carina.Subpetiolar process absent to moderately developed, if present, its apex forms a right angle.

Differential diagnosis
This species cannot be confused with other Malagasy Ponera species due to its dark brown color in contrast to the yellowish brown color of P. petila and P. swezeyi and its larger body size relative to the latter two species.Worldwide, this species most resembles the Indo-Australian P. clavicornis Emery (1900), but P. adumbrans sp.n. clearly differs from its congener in shape, color, and size characteristics.The two species exhibit non-overlapping ranges of morphometric ratios: P. adumbrans has a shorter head (cephalic index, CI: 77 [75, 78]) than P. clavicornis (CI: 81-85, see Taylor, 1967), and the scape of P. adumbrans is longer (SI: 93 [90,96]) than that of P. clavicornis (SI: 80-89, see Taylor, 1967).

Distribution
Known only from the Seychelles islands Silhouette and Mahé.

Differential diagnosis
In the Malagasy region, P. petila can only be confused with P. swezeyi; relevant diagnostic characters are discussed under diagnosis of the latter species.Worldwide, P. petila workers differ in size from all other species but P. szentivanyi Wilson, (1957).The latter species is reported to have a wider head (CWb: 340 µm, see Wilson, 1957) than P. petila (CWb: 333 µm [323,339]), but the scape index (SI) of the P. szentivanyi is considerably larger (SI: 94) than that of P. petila (SI: 85 [82,88]).

Distribution
In the Malagasy region, this species is distributed in Mauritius and Seychelles.Outside the region, it is known from Papua New Guinea and the Society Islands (French Polynesia).

Differential diagnosis
In the Malagasy region, this species can only be confused with P. petila, but petiole characters help distinguish these two species: in P. swezeyi, the anterolateral and posterior surfaces of petiolar node meet in sharp vertical ridges, while the lateral and posterior petiolar surfaces of P. petila meet in a rounded transition.Size of workers may also provide clues for identification: head width (CWb) of P. swezeyi is smaller 306 µm [291,323]) than that of P. petila (333 µm [323, 339]).Worldwide, P. swezeyi is the smallest species within the tenuis group.Size variance detected in a relatively large pool (n = 105) of P. swezeyi workers was found negligible (CWb: 291 µm to 323 µm) allowing for separation of P. swezeyi workers from that of other species.Only one species, P. szaboi Wilson, 1957) overlapping with P. swezeyi (CWb: 306 µm [291,323]), but the petiole of P. swezeyi species is considerably wider (PNI: 77-83) than that of P. szaboi (PNI: 61-65, see Taylor, 1967).

Distribution
Ponera swezeyi is known to occur in the eastern coast of Madagascar, Comoros, Mauritius, Mayotte, Reunion, and Seychelles in the Malagasy region.Moreover, populations are known from Queensland (northeastern Australia), Hawaii, Christmas Island, Japan (Fiji Island), Hawaii, Japan (Ogasawara Islands), Samoa, Solomon Islands, and Tanzania (Fisher & Bolton, 2016;Leong et al., 2019).
samples used in this study comply with the regulations for export and exchange of research samples outlined in the Convention on Biological Diversity and the Convention on International Trade in Endangered Species of Wild Fauna and Flora.For fieldwork conducted in Madagascar, permits to research, collect, and export ants were obtained from the Ministry of Environment and Forest as part of an ongoing collaboration between the California Academy of Sciences and the Ministry of Environment and Forest, Madagascar National Parks and Parc Botanique et Zoologique de Tsimbazaza.Approval Numbers: N° 0142N.EA03.MG02, N° 340N-EV10.MG04, N° 69 du 07.04.06,N° 065N-EA05.MG11, N° 047N-EA05.MG11, N° 083N-A03.MG05, N° 206 MINEN-VEF.SG.DGEF.DPB.SCBLF, N° 0324N.EA12.MG03, N° 100 l\fEF.SG.DGEF.DADF.SCBF, N° 0379N.EA11.MG02, N° 200N.EA05.MG02.Authorization for export was provided by the Director of Natural Resources.For research in the Seychelles, permits were obtained from the Seychelles Bureau of Standards, Ministry of Environment, and National Park Authority.Material Transfer Agreement was provided by the Seychelles Ministry of Environment.

Fig. 3
Fig. 3 A-C Ponera adumbrans sp.n. paratype worker (CASENT0159966).Head in full-face view (A), dorsal view of the body (B), lateral view of the body (C) Fig. 4 A-C Ponera petila non-type worker (CASENT0059796).Head in full-face view (A), dorsal view of the body (B), lateral view of the body (C)

Table 1
Abbreviations (Abbr.)formorphometriccharacters,and the protocol used to measure traitsAbbr.Measurement protocolCLCephalic length; maximum median length of head capsule.The head must be carefully tilted so the maximum length is positioned in the measuring plane.CWb Maximum cephalic width in full-face view.Measurement of the real cuticular surface and not of the diffuse pubescence surface.CS Absolute cephalic size.The arithmetic mean of CL and CWb.FRS Maximum width of frontal lobes.ML Diagonal length of the mesosoma in profile.Measured in lateral view from the anteriormost point of the anterior pronotal slope to the caudalmost point of the lateral metapleural lobe.MW Maximum width of pronotum.NOH Petiole node height; measured in a right angle from a reference line beginning at the transition point between the caudal node profile and the caudal petiolar neck and ending at the most dorsal point of node corner.PEH Maximum height of the petiole; measured from the subpetiolar toothlike process to the top of the petiole.PEL Petiole length; horizontal distance from the tip of the frontolateral node corner to the caudalmost point of the petiole.PEW Maximum width of the petiole in dorsal view.SL Maximum straight-line scape length excluding the articular condyle.

Table 2
Mean of morphometric ratios calculated for each species on individual level.Morphometric trait ratios and ± SD are provided in the upper row, and minimum and maximum values are given in parentheses in the lower row Mesosoma Pronotal-mesonotal articulation present and moderately developed across dorsum of mesosoma; sometimes a weakly defined shallow groove visibly interrupts the surface.Metanotal groove inconspicuous, rarely absent.Dorsal surface of mesosoma punctate, slightly shiny.Lateral surface of pronotum finely punctate, ventral area smooth and shiny.Mesopleural sculpture inconspicuous, medial and ventral part smooth and shiny; metapleural sculpture inconspicuously punctate, centrally smooth and shiny.Petiole Petiolar node not squamiform, moderately large; petiole width vs. absolute cephalic size (PEW/CS): 0.47 [0.43, 0.51]; slightly broader than long, (PEW/PEL): 1.20 [1.08, 1.35]; anterior face of node bluntly rounded, the sides usually slightly divergent posteriorly.Petiole node in profile relatively low and long, with a moderately long and weakly convex dorsum.Dorsal region of petiole smooth to finely punctate, shiny.Anterior and posterior faces of node usually weakly convergent dorsally, sometimes nearly parallel.Posterior surface of node bearing cuticular ridge or a fine vertical carina demarcating the posterior and lateral surfaces.Subpetiolar process well developed, its apex forms an acute angle.