An integrative taxonomic and phylogenetic approach reveals a new Neotropical swarm-founding social wasp, Pseudopolybia cryptica sp. n. (Vespidae: Polistinae: Epiponini)

Phenotypic characters are traditionally the main information for species discrimination in taxonomic studies of invertebrates. However, the presence of interand intraspecific polymorphism makes it difficult to identify species in many groups such as Neotropical social wasps. Herein, we examined different sources of biological information such as adult morphology, male genitalia, nest architecture, and genetic data applying an integrative taxonomic approach to study pinned museum specimens belonging to the social wasp genus Pseudopolybia de Saussure, 1863. Based on multiple independent lines of evidence, we described a new Neotropical swarm-founding social wasp, Pseudopolybia cryptica sp. n. Moreover, we proposed a phylogenetic hypothesis for Pseudopolybia including this new species. Our taxonomic findings applying an integrative approach reinforce that the social wasp diversity in the Neotropics may be underestimated due to morphological similarity.


Introduction
Accurate species delimitation is essential for multiple disciplines such as ecology, evolution, conservation biology, among others (Adams et al. 2014). However, species delimitation between closely related species is often difficult since traditional taxonomy depends on the evolution of informative and consistent phenotypic characters (Bickford et al. 2007;Gokhman 2018;Matos-Maraví et al. 2019). Social wasps (Hymenoptera: Vespidae) exemplify this issue since they show different sibling species mor pho logically similar (Menezes et al. 2011;Buck et al. 2012;Lopes and Menezes 2017;Schmid-Egger et al. 2017). Hence, as in other taxonomic groups, the combination of different sources of biological information such as morphology and genetics is necessary to better understand the diversity of social wasps. Pseudopolybia de Saussure, 1863 is a Neotropical genus of swarm-founding social wasps (Vespidae: Polistinae: Epiponini) comprising four species: Pseudopolybia compressa (de Saussure, 1854), Pseudopolybia difficilis (Ducke, 1905), Pseudopolybia langi Bequaert, 1944, andPseudopolybia vespiceps (de Saussure, 1863). They are distinguished by the third segment of the labial palpi bearing a short, stout, curved bristle near its apex, and the number of palpal segments being six maxillary and four labial (Richards 1978;Andena et al. 2007). The genus distribution is from Nicaragua to southern Brazil, and they are found mainly in rainforests (Richards and Richards 1951;Richards 1978;West-Eberhard et al. 1995). Nests of P. compressa, P. difficilis, and P. vespiceps have been described (Wenzel 1998). They are usually arboreal with spherical or tapering envelope and the entrance hole is a simple hole at the lowest point of the envelope (Richards 1978).
In the present paper, we use an integrative taxonomic approach, i.e. analyzing the variation of mitochondrial genetic markers with molecular species delimitation methods, adult morphology, nest architecture, and male genitalia, to assess the taxonomic status of pinned museum specimens previously identified as P. compressa from Central-West and Northeast of Brazil. Based on multi-ple independent lines of evidence, we described a new Neotropical swarm-founding social wasp, Pseudopolybia cryptica sp. n.  Bequaert (1944) and Smith (1857), respectively). For adult morphological structures and male genitalia we used the terminology following the original descriptions, Richards (1978), Andena et al. (2007), and Somavilla et al. (2018). The procedure for extraction and clearing of male genitalia followed the protocol proposed by Somavilla et al. (2018). We separated for study and description the following male genitalia structures: paramere, aedeagus, digitus, and cuspis. All observations and photographs of adult morphological structures and male genitalia were performed through a digital camera Leica DMC4500 coupled to a stereomicroscope Leica M205A and also using the Leica Application Suite software v4.10.0. A nest collected in Nova Mutum (Mato Grosso, Brazil) was photographed by SM, in situ, with a film camera and we edited the photographs using Adobe Photoshop CS6 Extended v13.

Methods
Additionally, we produced a distribution map for Pseudopolybia species based on a compilation of literature and collection records from museums (see Supplementary material S1). We georeferenced the localities from each specimen using the on-line Google Maps Platform. We excluded missing or inaccurate locality data, but the exception was a P. vespiceps specimen from Santa Catarina. The geographical coordinates were converted to decimal degrees and imported into a free and opensource geographic information system, QUANTUM-GIS v2.18.18 (Open Source Geospatial Foundation Project, Beaverton, OR, USA), to produce a distribution map.

Sanger sequencing was conducted by Centro de Recursos
Biológicos e Biologia Genômica (CREBIO), Universidade Estadual Paulista 'Júlio de Mesquita Filho' (Jaboticabal, São Paulo, Brazil). We carried out sequence quality control and assembly using the program GENEIOUS R7. We retrieved from GenBank nine DNA sequences (see Supplementary Table S1). All new DNA sequences were deposited in GenBank (accession numbers MT740709-MT740713 and MT738236-MT738241). Sequences were aligned using the default parameters of ClustalW (Thomp son et al. 1994) implemented in MEGA v7.0.26 (Kumar et al. 2016). All alignments were visually inspected and corrected. The most appropriate model of nucleotide evolution and the best-fitting partitioning scheme were selected using PartitionFinder2 (Lanfear et al. 2017) under the Bayesian information criterion (see Supplementary Table S2). We also computed a Kimura Two-Parameter (K2P) pairwise distances using MEGA v7.0.26 (Supplementary Table S3). Phylogenetic inference was conducted by maximum likelihood using IQ-TREE (ver. 1.6.12, see http://www. cibiv.at/software/iqtree; Nguyen et al. 2015) with a combined morphological and molecular dataset (concatenated COI and 16S sequences). We used the morphological matrix provided by Andena et al. (2007) with few modifications: we excluded the characters 2, 3, 13, 15, and 23, because they were not informative for our analysis, and adapted the character states of the characters 11, 17, and 21 (Supplementary Table S4). We used Agelaia pallipes (Olivier, 1792), Angiopolybia pallens (Lepeletier, 1836), and Apoica pallens (Fabricius, 1804) as outgroups based on previous phylogenetic studies (Andena et al. 2007;Menezes et al. 2020). Finally, we used the Jukes-Cantor type model for morphological data with ascertainment bias correction (MK+ASC) and calculated branch supports with 1000 replicates of ultrafast bootstrap approximation (UFboot) (Minh et al. 2013) and SH-like approximate likelihood ratio test (SH-aLRT) (Guindon et al. 2010) as implemented in the IQ-tree software. The consensus tree was visualized and edited using the program FigTree v1.4.0.
We carried out single-locus molecular species delimitation analyses with three methods based on different approaches: a sequence similarity clustering method, the automatic barcode gap discovery (ABGD) (Puillandre et al. 2012); and two tree-based coalescence methods, the Bayesian Poisson Tree Processes (bPTP) (Zhang et al. 2013) and the multi-rate Poisson Tree Processes (mPTP) (Kapli et al. 2017). The ABGD analyses were performed on the web-based interface http://wwwabi.snv.jussieu.fr/ public/abgd/ (Puillandre et al. 2012), with default values for both the proxy of the minimum relative gap width (X = 1.5) and the scanned range of prior intraspecific divergence (Pmin-Pmax = 0.001-0.1). Moreover, we checked the robustness of the ABGD results by changing the parameter values one at a time, particularly by increasing Pmax to 0.2 and by decreasing/increasing X (to 1 and 2). The bPTP and mPTP analyses were performed through the web interfaces http://species.h-its.org/ptp/ and https:// mptp.h-its.org/#/tree, respectively. We used as input maximum likelihood COI and 16S trees and concatenated maximum likelihood tree (generated by MEGA v7.0.26), removing the outgroups and using 300000 Markov chain Monte Carlo (MCMC) generations.

Results
Based on our integrative taxonomic approach, we can hypothesize that Pseudopolybia specimens collected in the Brazilian states of Mato Grosso and Bahia belong to a new species that we describe below. We consider several morphological characters including male genitalia ( Fig.  1 and Table 1), nest architecture ( Fig. 2), and two mitochondrial markers, COI and 16S, under a phylogenetic and molecular species delimitation approach (Fig. 3). Description. Female (Fig. 1A, B, C). Size: 13.0 mm. Forewing in length 9.5 mm. Head: (1) clypeus about 1.3 times wider than long, evenly convex in medial portion with lateral lobe flat; in contact with the eyes by less than the width of the antennal socket; lateral margins of the clypeus and upper margin sinuous; upper margin separated by less than the width of the antennal socket; apex of clypeus acute; entire surface of clypeus silver pubescent with yellowish and long bristles in apical portion;  punctures very shallow and spaced; (2) frons and vertex with long and spaced bristles and yellowish pubescence; punctures shallow, medium sized, separated by approximately one diameter; (3) eyes with middle bristles and spaced; (4) malar space with half of the antennal socket;

Pseudopolybia cryptica
(5) mandible with approximately 2.3 times longer than wide; with a long bristle band in the lower region; (6) middle region of the gena with width smaller than the eyes in profile; silver pubescence evident; middle bristles spaced; punctures very shallow and spaced; (7) diameter of the medial ocellus, 0,24 mm; (8) interocelar distance, 0,22 mm; (9) posterior region of head without occipital carena. Mesosoma: (1) pronotum with short and dense pubescence, concentrated on lateral part; pronotal carina produced high, lamellated with long bristles in front; punctures shallow, separated by less than one diameter; (2) Mesepisternum with dorsal groove present, at least as anterior trace with the same pattern of punctuation but becoming sparser laterally; short and dense pubescence; scrobal furrow wide, shallow; (3) tegula globose; (4) scutum with dense pubescence, central area more shinning; punctuation small, shallow, separated by one diameter or more, becoming sparser centrally; with a thin line in the upper central region; (5) scutellum with the same pattern of punctuation in the scutum, with a thin line in the entire central region; (6) propodeum with dense silver pubescent; long yellowish bristles laterally; posterior margin of the propodeum not elevated, muscle propodeal valve not translucid; (7) propodeal muscle large. Metasoma: (1) Tergum I compressed, cap-shape; punctures very weak, spaced; short and dense goldenish pubescence; (2) tergum II wider than long; punctures very weak, spaced; short and dense goldenish pubescence, row of short bristles; (3) tergum III-VI punctures pubescence like tergum II; (4) sternum II-V with punctures very weak; short and dense goldenish pubescence. Color: Black, mandibles, and malar space is blackish and yellowish; the lateral and apical part of clypeus, inner orbits, two spots on base of antennal scape, and part of antennal pedicel yellowish; flagellomere brownish; gena with a small yellow band, sometimes reaching the gena basis. Scutum, scutellum, metanotum, and mesepisternum blackish; margin of pronotum in dorsal view, yellow; tegula black; legs black to dark-brown in the portion of tibiae and tarsi; wings hyaline and venation dark-brown; tergum I-VI and sternum dark brownish and without apical yellow bands. male (Fig. 1D). Size: 11.0 mm. Forewing in length 8.5 mm. Like female except that the gena and clypeus are narrower and entirely yellow; the mandible is blackish and yellowish; the clypeus is covered with a long and dense golden pubescence; the frons is also yellowish in the interantennal area and antennal pedicel; small yellowish marks are present in the pronotal carina, end of tegula, anterior mesepisternum region, beneath of the coxae and femur; legs brownish. Male genitalia (Fig. 1E): (1) Paramere more long than wide, approximately two times longer than wide at the middle, basal angle obtuse, apical angle broad, long and curved spine very pointed apically, with long and dense bristles, mainly on spine base; (2) aedeagus apical portion with intermediate denticulation, extended on the apical portion to the end of the median expansion, an area near serration darker, lateral margin straight, penis valve widely dilated and with a slight central entrance, ventral process rounded and dilated in the sides, forming a "U", lateral apodema without a projection in the central curvature; (3) digitus pointed apically, with long and spaced hairs, punctation strong and forming a central band around the base of the digitus; (4) cuspis very pointed and tapering abruptly apically, with small and spaced hairs.
Nest (Fig. 2). The single arboreal nest of the new species used for nest's architecture description was collected by SM in the Buriti Farm, Nova Mutum, Mato Grosso, Brazil. It was in a rubber plantation close to a small lake. The spherical nest form flattened laterally was built around five meters from the ground in a rubber tree with thin branches incorporated in the upper part of the nest, and it measured approximately 42 cm in length and 27 cm wide ( Fig. 2A and B). The nest's envelope is predominantly gray (Fig. 2B) made with long fiber, typical of supple paper (Wenzel 1998). The external envelope is composed of many layers of sheets, and in the upper part of the nest is very hard and the lateral downward to the entrance is soft. The fourteen spherical multipedicellate combs stacked downward, growing gradually at margins sometimes in contact with envelope, cocoon with simple caps (Fig. 2C, E and F). The simple entrance hole at the bottom usually has than more one layer (Fig. 2D).

Diagnosis.
Color black, with few yellow marks mainly on the face and mandibles, clypeus, inner orbits, base of antennal scape, antennal pedicel and gena; clypeus wider than long, in contact with the eyes by less than the width of the antennal socket; frontal furrow deep; tegula globose; tergum I-VI dark brownish; sternum dark brownish; tergum and sternum without apical yellow bands.
Etymology. The specific epithet "cryptica" refers to the Greek word kryptikos that means hidden because the species was previously classified as Pseudopolybia compressa.
Distribution. Approximate range of the new species is shown in Fig. 4

Phylogenetic position.
The new species is reconstructed as sister to P. compressa based on the analysis of partial sequences of COI and 16S and morphological characters (Fig. 3). The detected level of genetic divergence between the new species and other congeneric range from 8.29-18.44% for COI and 3.34-11.08% for 16S (Supplementary Table S3).

Key to Pseudopolybia species
The following key is a revised and adapted version with few modifications of the key provided by Andena et al. (2007). Thus, P. cryptica would run down to P. compressa and P. vespiceps (couplet 3) and the following changes could be made to include the new species.  (Ducke) (Fig. 5B) 2' The first and second metasomal segment approximately twice as wide as long; metanotum slightly to strongly compressed; dark yellow with three black stripes on the scutum or blackish species with pale marks ................ 3 3 Dark yellow species with brown to blackish marks; prestigma much longer than wide; clypeus narrowly separated from eyes; eyes bare; the tip of the clypeus weakly pointed; metanotum slightly compressed; the malar space longer than fourth antennal article ..  (Fig. 1).

Phylogenetic reconstruction and molecular species delimitation analyses
The phylogenetic inference using a concatenated matrix (morphology and mitochondrial DNA) is summarized in Fig. 3. Pseudopolybia is supported as a monophyletic group. Pseudopolybia langi is sister to the clade with all other species (100% UFboot and SH-aLRT) and P. difficilis is sister to a clade composed of P. vespiceps, P. cryptica, and P. compressa (93% UFboot and 95% SH-aLRT). Our phylogenetic tree is very similar to previously proposed phylogeny for the genus (see AndenA et al. 2007), but here we included the new species that was recovered with high support (99% UFboot and SH-aLRT) as sister of P. compressa. The evolutionary divergences of COI and 16S using K2P are given in Supplementary Table S3. The average genetic distance within Pseudopolybia for COI was 12.09% and 16S was 7.8%. Pseudopolybia cryptica differs from P. compressa by a genetic distance of more than 8% for COI and 3.3% for 16S. The intraspecific genetic distance for COI and 16S within P. compressa range from 0.0%-3.36% and 0.47%, respectively (Supplementary Table S3). The ABGD, bPTP, and mPTP methods consistently identify P. cryptica as a distinct hypothetical species and also the other Pseudopolybia species already described (see bars in Fig. 3).

Discussion
Our integrated evaluation of adult morphology, male genitalia, nest architecture, and results from molecular species delimitation support the new species P. cryptica and indicate a sister species relationship with P. compressa. From the morphological point of view, P. cryptica cannot be confused with any Pseudopolybia species, except for P. compressa (Fig. 5D). The blackish general color of P. cryptica is very similar to P. compressa, but P. cryptica does not present extensive yellow marks, mainly in the tergum I and II. Moreover, the metasoma of P. cryptica is entire brownish and not black like P. compressa. Despite P. compressa has color variants described as three "morphotypes" [P. compressa morph. compressa (de Saussure, 1854), P. compressa morph. luctuosa (Smith, 1857), and P. compressa morph. laticincta (Ducke, 1904)], we noticed that they are morphological different of P. cryptica. According to Smith (1857) and Richards (1978), the P. morph. luctuosa is smaller and black, with a few yellow marks, and it is very similar to the P. compressa morph. compressa and P. compressa morph. laticincta. Pseudopolybia cryptica is different, since it is darker than all other P. compressa morphs, and it is larger than the P. compressa morph. luctuosa. Additionally, the male genitalia was another important character in the diagnosis of this new species, since it has aedeagus with intermediate den-  Bequaert, 1944. B: Pseudopolybia difficilis (Ducke, 1905) lectotype. C: Pseudopolybia vespiceps (de Saussure, 1863) type. D: Pseudopolybia compressa (de Saussure, 1854) holotype. ticulation in apical portion and paramere with long and sparse bristles in the spine base as differences when compared with the male genitalia of P. compressa (Table 1).
Concerning our molecular data, the species delimitation analyses have been carried out with a clustering (ABGD) and tree-based (bPTP and mPTP) methods on two mitochondrial markers, COI and 16S. Empirical studies have shown that species delimitation methods based on single-locus data tend to under-and oversplit species (da Silva et al. 2017;Pentinsaari et al. 2017;Renner et al. 2017). Additionally, despite P. cryptica showed high genetic distance when compared with other congeners (e.g. more than 8% for COI), a species-delimitation decision considering genetic distance using only mitochondrial markers has to be taken with extreme caution (e.g. Mueller 2006;Waichert et al. 2019). First, in the Epiponini wasps queen number may vary from many to few during the colonial cycle (cyclic oligogyny) and such colony strategy can lead to differences between nuclear and mitochondrial genetic structure at microgeographi cal scales (Ross and Shoemaker 1997). Second, mitochondrial markers are supposed to be linked, non-recombining locus with maternal inheritance, and likely evolved under similar constraints. We are aware of the limitations caused by stochasticity of the coalescent process and gene flow for single-locus species delimitation methods as well as the use of only mitochondrial mar kers, but for our study, we take advantage of using these methods within an integrative taxonomic framework. For instance, our morphology-based analysis and the three molecular species delimitation methods with two mitochondrial markers consistently recognize P. cryptica as a distinct taxon and confirm that it cannot be assigned to any known described Pseudopolybia species including the three "morphotypes".
Despite several studies performed molecular species delimitation approaches to identify Hymenoptera species (Fernández-Flores et al. 2013;Schwarzfeld and Sperling 2015;Williams et al. 2015;Hurtado-Burillo et al. 2017;Waichert et al. 2019;Brasero et al. 2020;Sabadini et al. 2020), molecular methods have not previously been applied to alpha-taxonomic problems in Epiponini wasps. Notably, based on morphological and molecular results for Synoeca species, Menezes et al. (2015) and Lopes and Menezes (2017) argued that the social wasp diversity in the Neotropical region may be underestimated due to morphological similarity of these insects, and hence they recommended the combination of morphology, po pulation-level sampling, and genetics to systematic studies for the group. Our findings here reinforce the need of combining independent sources of biological data for the diversity study of social wasps.

Acknowledgements
We are thankful to Dr. Gavin Broad and Dr. Jaswinder Boparai for support in Natural History Museum (London). We also are grateful