Morphological and genetic evidence supports the separation of two Tapinoma ants (Formicidae, Dolichoderinae) from the Atlantic Forest biome

Abstract The taxonomic boundaries of many Neotropical ant species of the genus Tapinoma are still unclear. Tapinoma atriceps and T. atriceps breviscapum are two morphologically similar taxa which occur sympatrically in the southern Atlantic Forest of Brazil. Some characters such as the scape length and head shape suggest that these taxa may be different species. We used DNA analysis and morphological evidence, including scanning electron microscopy, to evaluate the taxonomic validity of these taxa. We found distinct morphological characteristics that allow separating them as two different species, Tapinoma atriceps and Tapinoma breviscapumstatus novo, and this decision is supported by the DNA results, where Tapinoma atriceps was recovered as a lineage independent of T. breviscapum.


Introduction
Scape length (SL): the maximum length of the scape excluding the basal constriction. Weber'S, length (WL): in lateral view of the mesosoma, greatest distance from the approximate inflection point, where the pronotum curves into the cervical shield, to the posterior basal angle of the metapleuron.
The syntype worker of Tapinoma atriceps was measured from high resolution photographs using the program ImageJ v. 1.3 (Schneider et al. 2012). In the results, the measurements are presented as the mean value, followed by the standard deviation, with the minimum and maximum values between parentheses. Morphological terminology for wings follows Yoshimura and Fisher (2011).

Photographic resources and distribution map
High-resolution photographs of the specimens were captured using a Leica MZ16 stereomicroscope with a Leica DFC 500 camera, and final images were generated with Leica LAS 3D viewer LAS Montage v. 4.7. Integument surface and pilosity were examined using scanning electron microscopy (SEM) images generated with a JEOL JSM 6360-LV microscope under low vacuum (12-18 Pa) and a voltage acceleration of 15kV. Figure  plates were designed with InkScape v. 0.92 (available at http://www.inkscape.org).
The distribution map of the species was made with Quantum GIS v. 3.8 (QGIS Development Team 2017) using locality records of the examined material. The coordinate system used was UTM WGS84. When available, geographic coordinates were taken from the labels, otherwise, the coordinates were estimated using Google Maps by choosing a central point from the cited locality. For the final map composition, we used a polygon of the Atlantic Forest from the World Wild Fund (Olson et al. 2001). Biology information was extracted from literature, field observations, and label data.

Designation of type specimens
Lectotypes of Tapinoma atriceps and Tapinoma breviscapum were designated by taking a worker from the syntype series of each of these taxa. By affixing a single specimen as the name-bearing type of T. atriceps and a single specimen as the name-bearing type of T. breviscapum (Art. 74, ICZN 1999), it "permanently deprives all other specimens that were formerly syntypes of that nominal taxon of the status of syntype; those specimens then become paralectotypes" (Art. 74.1.3, ICZN 1999).

Statistical analysis
For evaluating possible relationships between morphometric characters in the workers of both taxa, especially those associated with the head, we constructed bivariate graphs (e.g., HL vs SL). Considering that the length and width of the head or the length of the scape appear to show variability between the workers and queens of T. atriceps and T. atriceps breviscapum, we analyzed the variability of HL, HW, SL, and WL between these two taxa using a parametric or a non-parametric comparison test, depending on the results of the Normality test of the data. For the latter, each of these morphometric characters were analyzed with a Shapiro-Wilks test. For the worker data set (n = 44), only SL showed normality (W = 0.94, p = 0.0875, α = 0.05; Suppl. material 1: Table S1), while for queens only SL and HW showed normality (Suppl. material 1: Table S1), although this last result may be biased by the small number of samples (n = 10). None of the measurements in the males showed normality. The variability of these morphometric traits (i.e., nonoverlapping differences) in workers and queens were analyzed using the Student's, t-test (T) with different sample sizes and different variances at a significance level of α = 0.05. In measurements with no normality, the difference of the two samples was evaluated with a Wilcoxon signed rank test at a significance level of α = 0.05. In the latter case, statistically significant differences were never found for any of the castes. All statistical analyses were performed in InfoStat v. 2020 (Di Rienzo et al. 2020).
DNA extraction, amplification, and sequencing DNA was extracted, amplified, and sequenced from eight workers of T. atriceps from seven localities and one worker of T. a. breviscapum from one locality in the Serra do Cipó, Minas Gerais, which is the only colony we managed to collect. Unfortunately, all other studied samples of T. a. breviscapum were unsuitable for DNA extraction. DNA was extracted from entire specimens using a GenElute TM Blood Genomic Extraction Kit (Sigma-Aldrich, Darmstadt, Germany) following the kit instructions. From each sample one worker was conserved as a voucher (Table 1). Standard polymerase chain reaction (PCR) methods were used to amplify partial fragments of the mitochondrial gene Cytochrome c oxidase subunit I (COI), the nuclear genes Long-wavelength Rhodopsin (LW Rh) and wingless (Wg), and an exon-primed intron-crossing marker (EPIC). Primers can be found on Table 2. DNA amplification was performed to a final volume of 25 µL. The PCR conditions for the COI marker were: 94 °C for 2 min, followed by 32 cycles of 94 °C for 45 s, 45 °C for 45 s, and 72 °C for 1 min, then 72 °C for 5 min. PCR conditions for Wg: 95 °C for 5 min, followed by 35 cycles of 92 °C for 1 min, 58 °C for 1 min, and 70 °C for 2 min, then 72 °C for 6 min. PCR conditions for the LW Rh marker: 95 °C for 5 min, followed by 35 cycles of 94 °C for 1 min, 56 °C for 1 min, and 70 °C for 1 min, then 72 °C for 5 min. PCR conditions for EPIC: 95 °C for 5 min, followed by 35 cycles of 92 °C for 1 min, 60 °C for 1 min, 70 °C for 1 min, then 72 °C for 6 min. All the sequences generated in this study were deposited in GenBank and the accession numbers are listed in Table 1. Table 1. List of specimens used in phylogenetic reconstruction and haplotype network from molecular data. Geographic information for each of the samples is recorded. The GenBank codes of these specimens are also included.

Analysis of genetic data
Consensus sequences were obtained with Staden Package (Staden 1996). The intronic region of LW Rh (LW Rhi) was separated and treated as a different marker than the exonic sequences (LW Rhe). For each marker, the sequences were aligned with Muscle (Edgar 2004) and then calculated nucleotide composition and p-distance in Mega X (Kumar et al. 2018). Composition for COI was analyzed by constructing a haplotype network using TCS network (Clement et al. 2002) in PopART v. 1.7 software (Leigh and Bryant 2015). DNA sequences for the species Tapinoma opacum Wheeler & Mann, 1914 and T. melanocephalum (Fabricius, 1793) were downloaded from Genbank and used as outgroups (Suppl. material 2: Table S2). Of these taxa, T. melanocephalum was selected to root the phylogenetic tree as one phylogenetic analysis previous suggest that T. opacum and T. atriceps are nesting in a clade of Neotropical species which is sister to the Nearctic clade (T. sessile + T. schreiberi), while T. melanocephalum is phylogenetically distant from those clades (R.J. Guerrero unpublished data). For the phylogenetic analysis, each of the four aligned loci were analyzed separately in a Bayesian phylogenetic framework using MrBayes v. 3.2.6. Each of the three genes was divided by codon position (position 1 + 2 and 3), along with the LW Rh intron, which was treated as another partition for this gene. The partitions and the best substitution models used by MrBayes (Suppl. material 3: Table S3) were determined using Akaike information criterion (AIC) with Parti-tionFinder v. 2.1 (Lanfear et al. 2012). The concatenated alignment consisted of 2287 base pairs (bp) including the five markers. All phylogenetic analysis were performed with MrBayes through the CIPRES science gateway (Miller et al. 2010). The parameters of the Bayesian analysis consisted of two independent runs of ten million generations each, with four Markov chains sampled every 1000 generations (mcmc ngen = 10000000 relburnin=yes burninfrac=0.25 printfreq=1000 samplefreq=1000 nchains=4 savebrlens=yes; sump relburnin=yes; sumt relburnin=yes; contype=halfcompat;). Tracer v. 1.6 (Rambaut 2018) was used to visualize parameter estimates and ensure that all estimates converged prior to removing a burnin period of 1 × 10 6 generations. Convergence time among runs was determined as twice the number of generations it took the standard deviation of split frequencies to drop below 0.01. Worker diagnosis. Lateral margin of head in frontal view distinctly convex. Compound eye with 9 or 10 ommatidia along maximum diameter. Scape long (SI > 93). In profile, dorsal margin of propodeum forms distinct angle with propodeal declivity; dorsal margin short, about 1/4 length of declivitous margin (Fig. 4B).
Head in full-face view oval, longer than wide, lateral margin convex, posterior margin slightly convex to straight (Fig. 1A). Maxillary palp relatively filiform, long, extending posteriorly beyond half of head. Masticatory margin of mandible with one large apical tooth, followed by two smaller teeth, fourth tooth larger than third, and then followed by denticles. Anterior margin of clypeus slightly emarginate medially. Scape almost as long as HL or greater (SI >93), surpassing posterior margin of head by distance equal to or greater than pedicel. In lateral view, dorsal margin continuously convex; metanotal groove weakly impressed; propodeum in lateral view slightly below level of mesonotum. Integument weakly imbricate, with exception of smooth petiole. Body covered by short, appressed pubescence. Head (excluding clypeus), antenna, and mesosoma lacking erect setae; clypeus with 6 long setae. Gastric tergites bearing erect hairs near their posterior margins: 2 hairs on first tergite, 2-4 on second, 4-6 on third, and 6-10 on fourth. Head and gaster medium brown; antennae, mesosoma, legs and petiole whitish yellow; mesosoma with brown spot on mesopleuron, spot sometimes present on lateroposterior corners of pronotum, metapleuron, and sides of propodeum.
Head subquadrate in full-face view, slightly longer than broad (CI 91-97), lateral margin very convex, posterior margin straight to slightly convex (Fig. 2B). Mandibular masticatory margin with one large apical tooth, followed by 2 smaller teeth, fourth tooth larger than third, followed by 5 smaller teeth, and then small denticles decreasing in size. Anterior margin of clypeus slightly emarginate medially. Scape relatively long, reaching or surpassing posterior margin of head by length shorter than that of pedicel (SI 82-83). Forewing with crossveins 2r-rs, 2rs-m, and cu-a present (Fig. 2C, D). Hindwing with cu-a present, cubitus short, not projected after 1rs-m+M. Integument weakly imbricate; mesopleuron smooth. Body covered by short yellowish pilosity, excepting glabrous petiole. Clypeus bearing 6 long setae; gastric tergites each bearing several erect setae near their posterior margins. Body color medium brown; palps, flagella, coxae, trochanter, tibiae, tarsi, and petiole whitish yellow; propodeum usually brown, but sometimes with whitish-yellow spot on posterodorsal region. Gastric tergites I-III with pale-yellow, posterior transverse strip. Head rounded in dorsal view; posterior margin slightly interrupted by posterior ocelli; anterior margin of clypeus straight to weakly emarginate medially. Eye large, rounded. Scape long, reaching or surpassing posterior head margin. Mandible semifalcate; masticatory margin with large apical tooth followed by denticles of similar size forming a serrated surface continuing indistinctly up to the mandibular basal margin. Integument feebly imbricate, katepisternum smooth. On forewing, 1m-cu absent, median short. On hindwing, free section of radial and cu-a present, free section of cubitus absent. Row of long setae present on posterior margin of fore and hindwings. Head, scutum, and gaster covered by moderate, yellow, short, appressed hairs; scutellum glabrous. Hairs absent to scarce on pronotum, mesopleuron, propodeum, and petiole. Antenna covered by short, decumbent hairs. Gastric tergites I-V lacking erect setae. Head, mesosoma, petiole, and gaster dark brown. Antenna and legs light brown.
Distribution. Tapinoma atriceps occurs in Argentina, Brazil, and Paraguay (Fig. 5). In Argentina, this species is present in the northeastern corner of the country, in the province of Misiones. In Brazil, our records show the presence of this species in the   states of Mato Grosso do Sul, Minas Gerais, Paraná, Rio Grande do Sul, Santa Catarina, and São Paulo. In Paraguay, T. atriceps occurs in the department of Canindeyú.
Biology. Tapinoma atriceps is an arboreal ant which can be found from the understory layer to the canopy and rarely on the ground. We found nests of this ant in hollow cavities of the vegetation or dry hanging branches, in plants of the families Poaceae (Bambusoideae), Melastomataceae, Piperaceae, and Urticaceae. Workers are commonly found foraging on the leaves of plants near the nest. The colony can be moderately large, with more than 312 workers, and in a couple of nests we found four dealate queens, evidencing polygyny as in other species of Tapinoma (e.g., Bustos and Cherix 1998;Buczkowski and Bennet 2008   Head in full-face view oval to rectangular, longer than wide; lateral and posterior margins slightly convex. Maxillary palps filiform, relatively short, not posteriorly surpassing beyond mid-length of head. Mandibles with masticatory margin with 1 large apical tooth, followed by 2 smaller teeth, fourth tooth larger than third, and then followed by denticles. Anterior margin of clypeus slightly emarginate medially. Scape relatively short when compared to T. atriceps (SI < 85), reaching or surpassing posterior margin of head by distance shorter than pedicel length. Pronotum and mesonotum form continuous feeble convexity in lateral view; metanotal groove weakly impressed; propodeum domeshaped, slightly higher than mesonotum. Integument weakly imbricate, excepting petiole which is smooth. Body covered with short decumbent pubescence. Head (excluding clypeus), antenna, and mesosoma lacking erect setae, clypeus with 6 anterior setae. Pilosity pattern on gastric tergites similar as to T. atriceps. Head and gaster medium brown; antenna, mesosoma, legs, and petiole pale whitish yellow to bright orange (Fig. 1C, D). Mesosoma with brown spot on mesopleuron and lateral pronotum, sometimes present on metapleuron and lateral propodeum, almost completely covering mesosomal side.
Head rectangular in full-face view, clearly longer than wide (CI 84-87); lateral and posterior margins straight. Masticatory margin of mandible with 1 large apical tooth, followed by 2 smaller teeth, fourth tooth larger than third, followed by 5 smaller teeth, and then small denticles decreasing in size. Clypeus slightly emarginate anteromedially. Scape short, never surpassing posterior margin of head (SI 72-76). Integument weakly imbricate, mesopleuron smooth. Dorsum of head with abundant, short, decumbent hairs; clypeus bearing 6 long hairs; gastric tergites with several erect setae near their posterior margins. Body medium brown; palps, flagellum, coxae, trochanters, tibiae, tarsi, and petiole whitish yellow to bright orange (Fig. 2E, F Head in dorsal view rounded, posterior margin slightly interrupted by lateral ocelli; anteromedian margin of clypeus straight to weakly emarginate. Compound eye large, rounded; scape long, reaching posterior margin of head; maxillary palp filiform. Mandible semi-falcate; masticatory margin with large apical tooth followed by many teeth of similar size. Integument feebly imbricate, katepisternum smooth. Forewing with median short; hindwing with free section of radial and cu-a present, free section of cubitus absent. Row of long setae present on posterior margins of fore and hindwing. Head, scutum, scutellum, and gaster covered by moderately abundant, yellow, short, decumbent hairs; antenna covered by short, decumbent hairs. Anepisternum covered by hairs, katepisternum lacking hairs ventrally. Gastric tergites I-V lacking erect setae. Head, mesosoma, petiole, and gaster dark brown; antenna and legs light brown. Distribution. Tapinoma breviscapum has been recorded from Misiones, Argentina, and from the Brazilian states of Minas Gerais, Rio de Janeiro, and São Paulo (Fig. 5).
Biology. Tapinoma breviscapum is an arboreal ant, but beyond that, there is not much available information. This species, reported as T. atriceps, was found inhabiting a gall of Microgramma squamulosa (Kaulf.) de la Sota (Santos et al. 2019

Morphological separation between Tapinoma atriceps and T. breviscapum
The most readily recognizable morphological diagnostic traits that permit separation of T. atriceps and T. breviscapum workers and queens are the relative length of the scape (i.e., SI), the shape of the propodeum, and differences in the degree of cephalic pubescence. In T. atriceps the worker scape is almost as long as the HL or greater (SI >93; Fig. 1A), in contrast with T. breviscapum, where it is relatively short (SI < 85; Fig. 1B), sometimes reaching or barely surpassing the posterior head margin by a distance shorter than the pedicel length. SL shows significant differences between the workers of each species (T = 7.51, p < 0.0001). Although there is a certain degree of overlap in the absolute measure (0.50-0.63 in T. atriceps and 0.44-0.56 in T. breviscapum) the relationship from SL to HL for each species showed non-overlapping ranges (Fig. 7). Other morphometric traits, such as HL, HW, and WL were also evaluated; however, each of their paired distributions overlapped, showing no statistical differences. The SL partially overlaps in queens of both species (0.58-0.62 in T. atriceps and 0.49-0.59 in T. breviscapum); however, differences between species were found (T = 2.29, p = 0.0257). These differences are notable in the non-overlapping ranges of the relative length of the scape (82-83 and 70-76, respectively). Statistical differences were also found in the HW of both species (T = 2.26, p = 0.0268); even without measuring, these differences are evident when they are compared under a stereoscope (Fig. 2B vs  2E), as T. breviscapum queens have a more elongate head as reflected in CI values that do not overlap those of T. atriceps queens.
The worker propodeum in both species differs markedly in shape and in the proportions between the dorsal and the posterior faces (Fig. 4B vs 6B). The dorsal propodeal margin when seen laterally in T. atriceps forms a distinct blunt angle with the declivity, contrasting with the rounded convexity formed in T. breviscapum. Additionally, the dorsal margin in T. atriceps is about 1/4 the length of the declivitous margin, while in T. breviscapum it is longer, about 1/2 that of the declivity (Figs 1B, D, 4B, 6B). The dorsal surface of the head in T. atriceps workers (Fig. 4A) is covered by appressed pubescence that is relatively longer and sparser than in T. breviscapum, where it is abundant and relatively shorter (Fig. 6A). The males of both species are relatively similar in morphology (Fig. 3), but the male of T. breviscapum can be differentiated from T. atriceps males because the former is on average slightly larger (0.63 ± 0.02 mm) and the scutellum is glabrous, while males of T. breviscapum are slightly smaller (0.62 ± 0.02 mm) and have decumbent hairs on the scutellum.

Genetic differentiation between Tapinoma atriceps and T. breviscapum
Final alignments had 648, 562, 422, and 655 bp for COI, LW Rh, Wg, and EPIC, respectively. For LW Rh, the length of the concatenated two flanking exonic sequences was 456 bp and for the intron 106 bp.
In the phylogenetic reconstruction (Fig. 8A), T. opacum is closer to T. atriceps and T. breviscapum than to T. melanocephalum. Tapinoma atriceps was recovered as a monophyletic group, sister to T. breviscapum. The Bayesian consensus trees of the individual analysis of each locus also recover both results (Suppl. material 4: Figure S1). Bayesian analysis of COI provided a topology similar to the Bayesian tree based on concatenated data, although with differences in branch lengths and node support of T. atriceps (PP = 0.71). The reconstructed separate trees with the nuclear loci also recovered T. breviscapum as a sister species to T. atriceps (Suppl. material 4: Figure S1) but the latter results in poor resolution among the sampled populations. Within T. atriceps, the topology derived from the concatenated data was relatively similar to the haplotype network (Fig. 8B). The samples from Paraná, corresponding to haplotypes H05 and H06, presented a comparatively deep divergence from the rest of the species. The sample of T. atriceps from Minas Gerais was sister to the group from southern Brazil, the populations of this latter group showing little genetic structure. Most of the nodes were relatively well-supported (PP 0.85-1.00) except for the low value of support corresponding to the Antonina haplotype (H01).
Seven mitochondrial (COI) haplotypes of Tapinoma atriceps were identified for the eight analyzed sequences and a single mitochondrial haplotype for the only T. breviscapum sample (Fig. 8B). The analysis estimated five unsampled haplotypes and found many mismatches between most of the haplotypes, evidencing high molecular variation for this marker. Haplotype H01, found in Paraná and Santa Catarina, is very close to haplotypes H02 and H03 and separated from them by only one nucleotide substitution; together, these form a group of haplotypes from southern Brazil. Other two haplotypes found in Paraná (H05 and H06) are relatively close to each other, and separated by one unsampled haplotype, but very different from the other haplotypes from Paraná. Haplotype H04 from Minas Gerais is closest to the H01-H03 haplotype group, separated by two unsam-   Figure S2) shows that the H07 haplotype is sister to the other haplotypes from southeastern and southern Brazil, which is congruent with the phylogenetic structure inferred from the concatenated molecular data (Fig. 8A).

Discussion
Tapinoma atriceps and Tapinoma breviscapum can be differentiated from other Neotropical Tapinoma ants by their particular bicolored pattern. Other Tapinoma can be mostly pale yellow or uniform brown, with yellow antennal scapes and coxae, but never with a spot on the mesopleuron, nor the bicolored pattern of T. atriceps and T. breviscapum. Only two other ant species that occur in South America, Tapinoma melanocephalum (Fabricius, 1793) and Linepithema leucomelas (Emery, 1894), have similar colors and size that could lead to confusion. In the case of T. melanocephalum, a common invasive species, the head and mesosoma is dark brown and the gaster is pale yellow (Guerrero 2018). Linepithema leucomelas can be differentiated by the characters that define the genus: presence of a well-developed petiolar scale and mandibular dentition which presents teeth alternating with denticles (Wild 2007b). Tapinoma atriceps and T. breviscapum are typical representatives of the genus in the Atlantic Forest of southwestern Brazil. Because of their sympatric distribution and morphological similarity, it has been difficult to separate them and the name T. atriceps has prevailed. Before the present work, T. breviscapum was only known from the type locality, Raiz da Serra in São Paulo (Forel 1908). We found that this species has a broader distribution, occurring in other Brazilian states such as Minas Gerais and Rio de Janeiro, and as far south as Misiones, Argentina. Such a broad distribution raises questions about the geographical origin of both species, which must be analyzed through a robust phylogeny and biogeographic inference of the genus Tapinoma in the Neotropical region.
Morphologically, T. atriceps and T. breviscapum can be differentiated by metric features associated with the head of the worker and queen, while the shape of the propodeum and hairs on the head allows the separation between the workers of both species, but the color of the body of the workers and queens in both species is relatively similar (see taxonomic treatment). Although the latter is true when species are allopatrically distributed, when occurring sympatrically they may exhibit no overlap of this trait (i.e., perhaps evidencing character displacement). The coloration pattern of workers and queens of T. atriceps and T. breviscapum from Serra do Cipó (MG, Brazil), corresponding to haplotypes H04 and H08, respectively, contrasts notably: antenna, mesosoma, legs, and petiole pale whitish-yellow in T. atriceps (Figs 1A, B, 2A, B) while those same sclerites are bright orange in T. breviscapum (Figs 1C, D, 2E, F). Differences in color are also observed in the worker and queen of both species from Misiones (Argentina), but the same sclerites which are bright orange in T. breviscapum are paler when compared to the H08 haplotype (R. Guerrero personal observation). This contrasting coloration pattern in sympatry could have played a fundamental role in the separation of lineages by reinforcing reproductive barriers between T. atriceps and T. breviscapum populations. The integration of comparative morphological analyzes of the genitalia of the males in both species and the analysis of more molecular data are necessary to elucidate aspects related to this evolutionary hypothesis.
The molecular analyses, including both mitochondrial and nuclear data, support the monophyly of T. atriceps (Fig. 8A), but we could not assess the monophyly of T. breviscapum because we could only analyze one sample of this species. The preliminary phylogenetic results of a broader study of Tapinoma, which includes several samples of both species from Minas Gerais and Misiones, confirm the reciprocal monophyly between them (R.J. Guerrero unpublished data). The COI-based Bayesian tree (Suppl. material 4: Figure S1) and the mitochondrial haplotype network (Fig. 8B) are very similar to the concatenated Bayesian tree, showing only minor differences in the position of the Misiones haplotype (H07) within T. atriceps. The other molecular markers also recovered T. breviscapum as sister to T. atriceps but failed to establish relationships among T. atriceps populations. Congruence between COI and the other nuclear markers is likely to be the result of similar differential lineage sorting. Although the Bayesian trees of COI and EPIC (Suppl. material 4: Figure S1) result in topologies with consistently different branch lengths, both markers show a similar phylogenetic relationship pattern within T. atriceps.
The average genetic distance between T. breviscapum and T. atriceps using COI (9.4%) is relatively high when compared with other Tapinoma species. For instance, Seifert et al. (2017) found genetic distances varying between 1.8% and 4.8% for pairs of Tapinoma species from the Mediterranean region using the same marker, considerably smaller values than those found in this study. The intraspecific variation in COI for T. atriceps is also considerably high (maximum of 9.6%) in comparison with T. ibericum Santschi, which has a distance of 1.3% as the greatest intraspecific variation (Seifert et al. 2017).
The highest values of intraspecific genetic distance in Tapinoma atriceps (9.6%) overlap with those between T. breviscapum and T. atriceps (8.8-10%); however, the greatest variation within T. atriceps species was found by comparing two samples from Paraná (H05-H06) with the rest of T. atriceps populations (Fig. 8B, Suppl. material 5: Figure S2). Such high genetic distance suggests the existence of cryptic diversity within this taxon, perhaps as the result of past climatic changes in the southern Atlantic Forest (Ströher et al. 2019), but divergence times estimates are necessary to obtain an approximation to this remarkable intrapopulation genetic differentiation. Despite the cryptic diversity suggested by COI we did not find any distinct morphological character in the workers and queens from Paraná (H05-H06) that would allow them to be separated from other T. atriceps specimens; therefore, we suggest these populations as part of the metapopulation of T. atriceps distributed in the Brazilian Atlantic Forest. Morphological analysis of more specimens, mainly males, in a wider geographical sampling throughout the Atlantic Forest biome could shed light on the intrapopulation mitochondrial genetic variation found in T. atriceps.

Conclusions
We found that a native Tapinoma occurring in the Atlantic Forest and previously considered as different phenotypes of the same species, correspond in fact to two different species, Tapinoma atriceps and Tapinoma breviscapum, based on morphological and molecular evidence. We also found high COI variation within T. atriceps populations, suggestive of cryptic diversity. However, these results should continue to be explored with a broader sampling, as more population samples might be needed to understand phylogeographic patterns in T. atriceps and T. breviscapum. Additionally, those phylogeographic patterns could help in understanding the biogeographic history of the Atlantic Forest. Finally, a complete phylogenetic framework is needed to understand the origin and evolution of Tapinoma in the Neotropical region.