A systematic review of the Neotropical social wasp genus Angiopolybia Araujo, 1946 (Hymenoptera: Vespidae): species delimitation, morphological diagnosis, and geographical distribution

For the Neotropical genus Angiopolybia Araujo 1946, several phenotypic forms were previously described, however, they have not been studied within an integrative taxonomic framework. Here, we used molecular data (variation of two mitochondrial genetic markers with molecular species delimitation methods) and morphology (adult morphology, male genitalia, and scanning electron microscopy images) to test the number of species within Angiopolybia. Specifically, we investigated the taxonomic validity of the morphological variants A. pallens dark morph, A. paraensis morph paraensis, A. paraensis morph ruficornis, and A. paraensis morph obscurior. Moreover, we reviewed the taxonomy and geographic distribution of the genus. Our results of morphological and molecular analyses are compatible with the current classification of Angiopolybia, and we did not find reasons to propose the morphological variants of A. pallens and A. paraensis as valid species. Additionally, we reassess the spatial range of the four Angiopolybia species and provide refined maps of their geographical distributions.


Introduction
Species boundaries are frequently hard to delimit due to intraspecific morphological variation. Hence, the use of different sources of biological information such as biogeography, behavior, ecology, molecular data, and morphology is generally regarded as a good practice for species delimitation (Bickford et al. 2007;Padial et al. 2010). Intraspecific polymorphism has been designated as "morph" for swarm-founding social wasps (Vespidae: Polistinae: Epiponini) (Richards 1978), but taxonomic investigations using a variety of characters from independent datasets are scarce for the group Lopes and Menezes 2017;Somavilla et al. 2021). Angiopolybia Araujo, 1946 is a neotropical swarmingfounding social wasp genus composed of four species: A. pallens (Lepeletier, 1836), A. paraensis (Spinola, 1851), A. obidensis (Ducke, 1904), and A. zischkai Richards, 1978. Angiopolybia is recovered as the sister lineage of all remaining Epiponini genera (Menezes et al. 2020;Noll et al. 2021). From a morphological perspective, Angiopolybia is characterized by the pronotum with lateral fovea, scutum with posterolateral lamella absent anteriorly and not adjacent to the tegula, mesoepisternum with dorsal groove, clypeus with square lateral lobes, and sharply pointed apex (Carpenter 2004). According to Andena et al. (2007), the monophyly of the genus is supported by six synapomorphies: prestigma about as long as wide, proepisternum with reduced carina, scutal lamella reduced, scutellum without an impressed line, practically straight dorsal groove, and flattened metapleural basalar area. Phylogenetic studies recovered two clades with the following relationship for Angiopolybia species: ((A. paraensis + A. obidensis) + (A. pallens + A. zischkai)) (Andena et al. 2007;Noll et al. 2021).
The genus occurs from Costa Rica to the south-central region of Brazil (Richards 1978), and A. pallens is the only species that occurs in the Amazon region and the Brazilian Atlantic Forest (Carvalho et al. 2014(Carvalho et al. , 2021. The nests of Angiopolybia are characterized as exposed on leaves, branches, and stone slabs and consisting of several stacked combs (Ducke 1914), fused or suspended from each other by a central pedicel, and the combs gradually grow along the margins (Wenzel 1998). The nests of Angiopolybia species have a single envelope, which can be ovoid shape or bottle-shaped, with a single entry at the lower part with a tubular form and horizontally curved (Richards 1978;Wenzel 1998). Morphological difference between queens and workers is subtle (Richards 1978), and only the male of A. pallens was described by Richards (1978).
Morphological differences between A. paraensis and A. obidensis and between A. pallens and A. zischkai are subtle, such as the height of the anterior pronotal lamella and the development of the pronotal lobe, respectively (Richards 1978;Andena et al. 2007). Moreover, Richards (1978) characterized multiple morphological variants of A. pallens and A. paraensis as "morph", such as A. pallens dark morph, A. paraensis morph paraensis, A. paraensis morph ruficornis, and A. paraensis morph obscurior. However, Andena et al. (2007) made the following comment about the "morphs" of A. paraensis: "is a category without nomenclatural standing, hence these morphs are synonyms of A. paraensis".
Here, we analyzed the variation of mitochondrial genetic markers with molecular species delimitation methods, adult morphology, male genitalia, and scanning electron microscopy images to characterize in detail the species limits within Angiopolybia. Thus, we tested whether the "morphs" of A. pallens and A. paraensis can be recognized based on morphological and molecular data, as well as providing diagnostic features for Angiopolybia species and a reassessment of their geographical distribution.

Morphological analysis
We used adult specimens for a redescription of the female and description of the male. We examined 469 females and 32 males of A. pallens (eight males for genitalia analysis), four females of A. zischkai, 104 females and three males of A. obidensis (two males for genitalia analysis), and 80 females and six males of A. paraensis (four males for genitalia analysis). We also analyzed images of five type specimens of the species. They were analyzed under a Nikon SMZ645 stereo microscope with an accessory magnifying lens of Nikon G-AL 2x. We obtained the proportions and morphological measurements with the ocular lens AmScope reticulated WF10X/22 and the measurements in the images using the IMAGEJ 1.52a (Rasband 2019). The explanation of how each measurement was performed is described in Supplementary Material (Table S1). For the morphological terminology, we followed Richards (1978), and for male terminalia, we used Richards (1978), Buck et al. (2012), andSomavilla et al. (2018). We obtained the images of specimens and morphological characters with the aid of a Leica DMC4500 digital camera coupled to a Leica M205A stereomicroscope with a self-assembly system, using the Leica Application Suite v.4.10.0-Montage® software. We edited the images and assembled the plates using the software ADOBE PHOTOSHOP® CS6 v.6.1. We employed scales of 1 mm and 0.5 mm for images of adults and male genitalia, respectively. We also generated Scanning Electron Microscopy images (equipment Oxford Instruments INCAx-act, model 51-ADD0007) without the use of preparation technique of coating. Finally, we built a matrix of morphological characters based on the studies of Ducke (1914), Richards (1943Richards ( , 1978, Carpenter (1991), and Andena et al. (2007), as well as proposing new morphological characters for Angiopolybia species.

Molecular dataset
We extracted total DNA from a mid and a hind leg of specimens preserved in ethanol and pinned museum specimens. We extracted DNA from a single specimen of A. pallens, A. zischkai, A. obidensis, A. paraensis yellow, A. paraensis brown and yellow and A. paraensis dark brown, the complete list of specimens used in our analyses is presented in Supplementary Material (Table S2). For extraction, we used the DNeasy® Blood & Tissue extraction kit (QIAGEN®, Valencia, California, USA) following the manufacturer's protocol. We amplified two mitochondrial gene fragments, Cytochrome Oxidase subunit I (Cox1) and 16S ribosomal DNA (16S). Specific primers and conditions for PCR amplification are described by Menezes et al. (2015Menezes et al. ( , 2017. The PCR products were purified using exonuclease I and shrimp alkaline phosphatase. Sanger sequencing was conducted by the Centro de Recursos Biológicos e Biologia Genômica (CREBIO), Universidade Estadual Paulista 'Júlio de Mesquita Filho' (UNESP, Jaboticabal, São Paulo, Brazil). The sequencing was carried out in both directions, and the consensus sequences were assembled on GENEIOUS R7 (Kearse et al. 2012). All sequences were deposited at GenBank (access numbers are MZ496308-MZ496314 and MZ513916 for Cox1, and MZ496381-MZ496387 and MZ513917 for 16S). We performed the alignment using the MUSCLE algorithm (Edgar 2004), implemented in the program MEGA X (Kumar et al. 2018) and with default parameters. We visually inspected and corrected all alignments. We concatenated the Cox1 and 16S sequences using the software MESQUITE v.3.61 (Maddison and Maddison 2019). The most appropriate model of nucleotide evolution and the best-fitting partitioning scheme were selected using PARTITIONFINDER2 (Lanfear et al. 2016) under the Bayesian information criterion (see Table S3).

Phylogenetic and molecular species delimitation analyses
Phylogenetic inference was conducted by Maximum Likelihood (ML) using the IQ-TREE v. 1.6.12 (Nguyen et al. 2015) with a concatenated morphological and molecular dataset. We calculated branch supports with 1000 replicates of SH-like approximate likelihood ratio test (SH-aLRT) (Guindon et al. 2010) and Ultrafast Bootstrap approximation (UFBoot) (Minh et al. 2013). We used Apoica thoracica du Buysson, 1906 and Agelaia fulvofasciata (DeGeer, 1773) as outgroups based on previous phylogenetic studies (Andena et al., 2007;Menezes et al. 2020;Noll et al. 2021). Visualization and edition of the phylogenetic trees were performed using the program FIGTREE v.1.4.2 (http://tree.bio.ed.ac.uk/software/ figtree). We also performed analysis under Maximum Parsimony (MP) with our concatenated matrix using the program TNT -Tree analysis using New Technology v.1.5 (Goloboff and Catalano 2016). We performed a Traditional search by Wagner trees with 1 random seed, 1000 replicates, saving 30 by replication, and collapsing the tree after the search. The polarization of states was carried out by comparison with outgroups (Nixon and Carpenter 1993). The visualization of the MP phylogenetic tree was performed using the program WINCLADA v.1.00. 08 (Nixon 2002). We used four species delimitation methods for Cox1 and 16S data separately and also concatenated as follows: Automatic Barcode Gap Discovery (ABGD) (Puillandre et al. 2012), Assemble Species by Automatic Partitioning (ASAP) (Puillandre et al. 2021), bayesian Poisson Tree Processes (bPTP) (Zhang et al. 2013) and multi-rate Poisson Tree Processes (mPTP) (Kapli et al. 2017). For ABGD and ASAP, we used the online platforms https:// bioinfo.mnhn.fr/abi/public/abgd and https://bioinfo. mnhn.fr/abi/public/asap respectively, using the Kimura Two-Parameter (K2P) model (Kimura 1980). We used the phylogenetic trees generated by IQ-Tree ver.1.6.12 for the bPTP and mPTP methods. For the bPTP and mPTP, we used the online platforms https://species.h-its.org and https://mptp.h-its.org/#/tree respectively, removing the outgroups and with default parameters, except for the number of generations of Monte Carlo Markov Chains (MCMC) of 300,000 in bPTP. Additionally, we analyzed the genetic distance among the specimens for both mitochondrial markers in MEGA X using the K2P model.

Geographic distribution maps
We used the recorded localities for Angiopolybia specimens to build a geographical database of species occur-rence. The recorded localities were based on the label data of all analyzed specimens and information obtained in the literature with the precise locality. We avoided the use of non-geo-located information, such as presence data in states or provinces. The Google Maps platform was used to determine each specimen's geographical coordinates and convert them to decimal degrees. We used the resulting database to characterize in detail the geographic range of each Angiopolybia species with the software QGIS v. 3.10.11 (QGIS.org, 2020, Geographic Information System, http://www.qgis.org 2020) and a biome map reported by Dinerstein et al. (2017).

Results
To characterize in detail the species limits within Angiopolybia, we considered several morphological characters, including females, males, male genitalia, and nest architecture , and two mitochondrial markers, Cox1 and 16S, under a phylogenetic and molecular species delimitation approach (Fig. 9). Additionally, we performed a taxonomic revision for Angiopolybia and characterized the geographical range of each species (Fig.  10). Moreover, we proposed an identification key with new morphological characters and images.  Ducke (1914), these genera are not well-defined by the proposed characters. Based on this, some morphological characters that had been proposed to separate Angiopolybia from Agelaia such as stelocyttarous nest (with envelope) (Ducke 1914), very weak or absent occipital carina (Richards 1943), and flat metapleural basalar area (Richards 1978;Andena et al. 2007) were not considered here because they are not exclusive to Angiopolybia, since they are present in few species of Agelaia. Moreover, the scutellum with line or depression (Richards 1943;Andena et al. 2007) was not considered because they are not exclusive to Agelaia, since some Angiopolybia species present the line, despite it is inconspicuous. Richards (1943) commented that the characters determined by him, isolated, had no effect on separating the taxa.

Revision of Angiopolybia
Based on previous and this study, A. pallens and A. zischkai are closer, as are A. paraensis and A. obidensis. Angiopolybia pallens and A. zischkai are distinguished by rounded gena and propodeum with posterior submedian translucent mark inserted in a depression, whereas the other species have angled gena and submedian translucent mark of the propodeum not inserted in a depression. We documented other differences in the step 1 of the identification key for the genus (see below). Stelopolybia (Angiopolybia) pallens; Richards and Richards, 1951: 77 (list and notes about the nests). Type locality. Cayenne, French Guiana.

Diagnosis.
Anterior wing of 7-8.5 mm; eyes with medium-sized and sparse bristles; rounded gena; pronotal lamella low on the anterior margin, one fifth of the height of antennal socket; pronotal lobe developed in the anterior lateral region, below the pronotal fovea; defined and deep pronotal fovea; axillary fossa with anterior margin directed to the posterior region; posterior submedian translucent mark of the propodeum inserted in a round depression; basal metapleural area with parallel upper and lower margins. (Fig. 1a,   (1) Metasomal tergum I two times longer than broad. Tergum with angulation in the posterior third, in lateral view. (2) Metasomal tergum II 0.82 times longer than broad. Color. Brown in general. Yellow: lateral of the apex, lateral of the front, interantennal region, disc of the clypeus, mandibles, gena, antennal segments: scape, dorsal of FL6 (flagellomere 6) and 7, FL8-10, band contouring the posterior margin of the pronotum, tegula, thin bands in the posterior margins of the metasomal terga I, II and III, and metasomal sterna II and III. Yellowish brown: central longitudinal and lateral region of the mesoscutum, tibiae and tarsi, femora median and posterior, metasomal tergum I, anterior half of metasomal tergum II, metasomal sterna I-IV. Black: FL1-5, ventral of FL 6 and 7, metasomal terga 4-6 and metasomal sterna 5 and 6. Wings with hyaline cells, pterostigma and venation in general yellowish-brown, except brown in the veins C, Sc+R, M+Cu and M. Fig. 1d, e, f). Size. (1)  (1) Metasomal tergum I 1.9 times longer than broad. (2) Metasomal tergum II 0.8 times longer than wide. Genitalia (Fig. 6a-e). Paramere 1.6 mm long and 0.6 mm wide; parameral spine with one fifth of the paramere, curved upwards and with small-sized and sparse bristles; lobe with a rounded apex and slightly curved downwards. Aedeagus 1.2 mm long; enlarged valve with a small emargination in the tip; apical portion 0.42 mm long and straight, ventral margin with denticles directed for the anterior region; denticulation decreasing in size from the apex to the base and more sclerotized than the rest of the apical portion; small-sized bristles with alveolar base, closer in the lower half and sparse in the upper half; median expansion without denticles and with acute apex; lateral apodeme not flattened dorsoventrally at the apex; basal apodeme arched to the venter. Digitus 2.5 times longer than wide; apical process not curved in the region of the upper half and with bristles of alveolar base small and sparse; rounded anteroventral lobe with a strip of black scale-like bristles crossing it obliquely at the base of the digitus; bristles absent in the basal articulation. Cuspis approximately 0.48 mm long, with five black scale-like bristles on the lateral lobe, and small bristles with alveolar base and close throughout the area of the cuspis, except sparse in the central region and on the ventral margin.

Redescription of male (
Morphological variation (Fig. 7a-g). Anterior wing between 7-8.5 mm in length. Anterior margin of the pronotum (below the fovea) more curved. Pronotum, in lateral view, with frontal region more projected forward. Coloration varied between populations from black with yellow marks [like A. pallens dark morph (sensu Richards, 1978)] to yellowish brown. Nest (Fig. 2a-c). The nest of A. pallens was initially described by Möbius (1856) for Polybia ampullaria (junior synonym of the taxon), and later described by Wenzel (1998) as a nest with a flask-shaped envelope and with a long downward entrance with its hole horizontally; pedicel initially single and later being able to be multiple; flexible card with long fibers, usually yellow or amber; and adjacent combs, suspended or fused from each other and without contact with the envelope. Additionally, the following nest variation was found: built on an irregular substrate (thin branch with several leaves), connected to the substrate by a central pedicel (thick) and several support pedicels (fines); 11 combs overlapping and also connected by a central pedicel and multiple support, with the third and fourth combs with the largest circumferences and combs decreasing in circumference towards the ends, and without contact with the envelope; hexagonal cells of the combs with diameter of 2.5 mm; single envelope with long fibers arranged longitudinally, circumference gradually decreasing towards the entrance and without entrance of tubular shape.
Comparative comments. Angiopolybia pallens is distinguished by the presence of a lobe in the lateroanterior region of the pronotum, absent in A. zischkai, and metapleural basalar area with parallel upper and lower margins, diverging in A. zischkai. Some A. pallens specimens can be confused with A. zischkai because it can resemble the typical form of A. zischkai, of color darker.
Additional comments. Despite the geographic disjunction by thousand kilometers in the distribution of A. pallens between the Amazon and Atlantic Forest (Carvalho et al. 2014(Carvalho et al. , 2021, considering the morphological analysis of specimens collected in both biomes, we verified that the differences, when found, are very subtle and follow the variation found between different populations. Considering the populations from the Atlantic Forest, the color varied from yellow with brown marks to completely brown, while among the populations from the Amazon Forest it varied from black with yellow marks to yellowish brown, with some forms existing in both biomes. The occipital carina present in A. pallens, despite weak or not, found in this study not reported by Richards (1978) and Andena et al. (2007). They treated the occipital carina absent for the taxon. The prestigma longer than wide, also, was not cited by Richards (1978) and Andena et al. (2007). However, this character was found by Silveira and Carpenter (1995) for some specimens, and here we also found the two forms, prestigma about as long as wide and longer than wide. The redescription of A. pallens was made based on the lectotype of Polybia infernalis de Saussure, 1854 (junior synonym of A. pallens). The information about the male specimen described is: BRA, Roraima, Amajarí, Serra do Tepequém, SESC Tepequém.

Angiopolybia zischkai Richards, 1978
Figs 3a-d, 7h Angiopolybia zischkai Richards, 1978: 30 (list  Type locality. Zumbi, Ecuador. Diagnosis. Anterior wing of 8-9.5 mm; eyes with medium-sized and sparse bristles; rounded gena; pronotal lamella low on the anterior margin, one fifth of the height of antennal socket; pronotal lobe not developed in the lateral anterior region, below of the pronotal fovea; defined pronotal fovea, but little deep; axillary fossa with anterior margin directed to the posterior region; posterior submedian translucent mark of the propodeum inserted in a round depression; metapleural basalar area with divergent upper and lower margins. (Fig. 3a-c). Size. (1) Head 1.03 mm long, 2.09 mm high, and 2.24 mm wide; (2) mesosoma 3.54 mm long, anterior wing 9.42 mm long, and posterior wing 6.10 mm long; (3)   deum, lateral margin of the propodeum, upper region of the metapleural basalar area. Yellowish brown: FL7-10 of the antenna, anterior and median coxae, trochanters, femora, tibiae and tarsi (but with dorsal brown spots). Black: ocellar area, FL1-6 of the antenna, mesoscutum and metasoma. Wings with hyaline cells, except yellow in the costal, medial, submarginal I and marginal; and yellowish-brown venation, except brown in the veins C, Sc+R, M+Cu, M and in the beginning of the Cu.

Male. Unknown.
Morphological variation (Fig. 7h). We found a specimen with yellow color and black marks, and the abdomen, apparently, wider in dorsal view.
Nest. Not described, but Valverde et al. (2019) in the identification key of social wasps from Costa Rica commented that the nest envelope resembles an inverted flask.
Comparative comments. Angiopolybia zischkai resembles A. pallens, but it is distinguished by the pronotum without a developed lobe in the lateroanterior region, which is present in A. pallens; metapleural basalar area with divergent upper and lower margins, which are parallel in A. pallens; and pronotal fovea with translucent mark of elliptical shape, which is circular in A. pallens.
Diagnosis. Anterior wing of 12-14 mm; eyes with very small-sized and sparse bristles; angulate gena with enlarged lower and upper region; pronotal lamella very elevated along the anterior margin, one third of the height of antennal socket; axillary fossa with anterior margin directed to the anterior region; posterior submedian translucent mark not inserted in a depression. (Fig. 4a, b, c). Size. (1) Head 1.77 mm long, 3.13 mm high, and 3.56 mm wide; (2) mesosoma 5.69 mm long, anterior wing 13.38 mm long, and posterior wing 8.55 mm long; (3) (6) Anterior wing with prestigma as long as wide. (7) Posterior wing with 11 hamuli. Metasoma.

Redescription of female
(1) Metasomal tergum I 1.7 times longer than broad. Tergum with angulation in the posterior third, in lateral view.
(2) Metasomal tergum II 0.8 times longer than broad and without a row of very long bristles on the posterior margin. Color. Brown in general. Yellow: median longitudinal band and lateral of the frons, interantennal elevation, band surrounding the disc of the clypeus, mandibles, lower quarter of the gena, malar space, band contouring the posterior margin of the pronotum, tegula, spot anterior to the scrobal furrow of the mesepisternum, longitudinal submedian band and thin lateral band in the mesoscutum, axilla, anterior half of the scutellum, metanotum, submedian band in the propodeum, margin anterior to the propodeal valve, upper region of the metapleural basalar area, apex of the coxae, femora, tibiae and tarsi, bands along the posterior margins of the metasomal terga I-III and sternum II. Yellowish-brown: vertex and gena. Black: metasomal terga II-VI and metasomal sterna II-VI. Reddish-brown: inferior margin of the clypeus and the mandibular teeth. Wings with yellowish-brown in the cells and venation, except reddish-brown in the veins C, Sc+R, M+Cu, M and in the beginning of the Cu. (Fig. 4d, e, f)  half of the width of the eye at the level of the ocular sinus. Mesosoma. (1) Mesoscutum 1.1 times longer than wide. Metasoma. (1) Metasomal tergum I 1.7 times longer than broad. (2) Metasomal tergum II 0.8 times longer than wide. Posterior margin with slight emargination in the center. Genitalia (Fig. 6f-j). Paramere 2 mm long and 0.81 mm wide; parameral spine with one fifth of the paramere, straight and with long bristles; lobe with rounded apex and slightly curved downwards. Aedeagus 1.72 mm long; enlarged valve with a small emargination in the tip; apical portion 0.66 mm long and curved to the venter, ventral margin with denticles directed for the anterior region; denticulation with large and conical denticles in the basal and apical thirds and small denticles with widened bases in the median third, more sclerotized than the rest of the apical portion; small-sized bristles with alveolar base, closer in the lower half and sparse in the upper half; median expansion with one denticle and with acute apex; lateral apodeme not flattened dorsoventrally at the apex; basal apodeme arched to the venter. Digitus 3.4 times longer than wide; apical process little curved in the region of the upper half and with bristles of alveolar base small and sparse; rounded anteroventral lobe with a strip of black scale-like bristles crossing it obliquely at the base of the digitus; bristles absent in the lower margin and basal articulation. Cuspis approximately 0.44 mm long, with 16 black scale-like bristles on the lateral lobe, and long bristles with alveolar base and close throughout the area of the cuspis, except sparse in the central region and on the ventral margin.

Description of male
Morphological variation (Fig. 7i). Some A. obidensis specimens found in São Gabriel da Cachoeira (Amazonas) and Parque Nacional Serra da Mocidade (Roraima) are darker, like A. paraensis of coloration transitional between the yellow form and the black and yellow forms, which also occurs in these regions.

Nest. Unknown.
Comparative comments. A. obidensis resembles A. paraensis, but it is distinguished by the lamella along the anterior margin of the pronotum which is very elevated (one third of the height of antennal socket) in A. obidensis, and low (one fifth of the height of antennal socket) in A. paraensis; pronotum with prominence absent in front of the fovea, but slight prominence in front of the fovea in A. paraensis; angulate gena with enlarged lower and upper region, but gena with only enlarged lower region in A. paraensis; parameral spine straight, but spine curved upwards in A. paraensis; aedeagus with lateral apodeme not flattened dorsoventrally at the apex, but apex of the lateral apodeme flattened in A. paraensis.
Additional comments. The label information of the male specimen described is Brazil, AM, Itacoatiara, Mil Madeireira. 16.xii.1999 (Fig. 4a, b, c). Type specimen analyzed by images.
Diagnose. Anterior wing of 13-15 mm; eyes with very small-sized and sparse bristles; angulate gena with enlarged lower region; pronotal lamella low on the anterior margin, one fifth of the height of antennal socket; axillary fossa with anterior margin directed to the anterior region; posterior submedian translucent mark not inserted in a depression. (Fig. 5a, b, c). Size. (1) Head 2.01 mm long, 3.35 mm high, and 3.71 mm wide; (2)  (1) Metasomal tergum I 1.5 times longer than broad. Tergum with angulation in the posterior third, in lateral view.

Redescription of female
(2) Metasomal tergum II 0.7 times longer than broad and with a row of very long bristles on the posterior margin. Color. Yellowish-brown in general. Yellow: lateral of the vertex, lateral of the frons, interantennal elevation, lateral and the lower margin of the clypeus, mandibles, gena, anterior half of the lower quarter of the gena, malar space, band contouring the posterior margin of the pronotum,  II-IV. Brown: a spot that extends from the vertex to the middle of the frons, transversal band in the pronotum, scrobal furrow, region posterior to the scrobal furrow, metapleural basalar area, lateral bands of the propodeum, and metasomal terga I and II. Dark brown: ocellar area, mesoscutum, posterior half of the scutellum, posterior margin of the metanotum, bands in furrow and in anterior marginal of the propodeum, and metasomal terga III and IV. Reddish-brown: lower margin of the clypeus and mandibular teeth. Reddish yellow antenna. Wings with yellowish-brown in cells and in venation, except reddish-brown in the veins C, Sc+R, M+Cu, M, at the beginning of the Cu and pterostigma.
Description of male (Fig. 5d, e, f). Size. (1) Head 1.40 mm long, 3.08 mm high, and 3.47 mm wide; (2) mesosoma 5.28 long, anterior wing 14.2 mm long, and posterior wing 9.9 mm long; (3)  (1) Metasomal tergum I 1.7 times longer than broad. (2) Metasomal tergum II 0.8 times longer than wide. Genitalia (Fig. 6k-o). Paramere 2.3 mm long and 0.9 mm wide; parameral spine with one fifth of the paramere, curved upwards and with long bristles; lobe with rounded apex and not curved. Aedeagus 1.68 mm long; enlarged valve with a small emargination in the tip; apical portion 0.72 mm long and curved to the venter, ventral margin with denticles directed for the anterior region; denticula- tion with large and conical denticles in the basal and apical thirds and denticles reduced in the middle third, more sclerotized than the rest of the apical portion; small-sized bristles with alveolar base, closer in the lower half and sparse in the upper half; median expansion without denticles and with acute apex; lateral apodeme flattened dorsoventrally at the apex; sinuous basal apodeme. Digitus 2.7 times longer than wide; apical process little curved in the region of the upper half and with bristles of alveolar base small and sparse; rounded anteroventral lobe with a strip of black scale-like bristles crossing it obliquely at the base of the digitus; bristles absent in the lower margin and in the basal articulation. Cuspis approximately 0.46 mm long, with 26 black scale-like bristles on the lateral lobe, and long bristles with alveolar base and close throughout the area of the cuspis, except sparse in the central region.
Morphological variation (Fig. 7j-l). Anterior wing of 13-15 mm; posterior wing of 13-19 hamuli; Angiopolybia paraensis occurs in three color variants, identified by Richards (1978) as A. paraensis morph paraensis (yellow specimens), A. paraensis morph ruficornis (brown and yellow specimens) and A. paraensis morph obscurior (brown specimens). Angiopolybia paraensis show small changes between yellow populations and between brown and yellow populations, such as a slightly darker color or some yellow marks, respectively, but they are well-defined. Despite the color variation, the morphological characteristics of female and male adults, and male genitalia used in the description are preserved in the three forms, so they should not be treated as subspecies or differentiated as morphs. Only a few A. paraensis populations, with dark brown color from French Guiana and Suriname, showed morphological variations such as the absence of very long bristles in the posterior margin of the metasomal tergum II and propodeum with a region anterior to the spiracle less projected in the metapleural basalar area. However, we believe that this evidence is still insufficient to justify a new species. Moreover, the analysis of species delimitation with molecular data (see below) showed that these variants are within the intraspecific limits of A. paraensis.
Nest. Described by Schulz (1903) as a spherical nest about 25 mm in diameter (possibly incorrect unit of measurement was used, with centimeter (cm) being the correct). Ducke (1910) complements the description of Schulz (1903), stating that the nest is composed of four overlapping combs, joined by a central pedicel, with a simple and very resistant envelope, and with transverse streaked with light and dark colors.
Comparative comments. Angiopolybia paraensis resembles A. obidensis, but it is distinguished by the pro-notum with low lamella in the anterior margin (one fifth of the height of antennal socket), being high lamella (one third of the height of antennal socket) in A. obidensis; pronotum with a slight prominence in front of the fovea, but prominence absent in A. obidensis; gena not enlarged in the upper region, but enlarged in the upper region in A. obidensis; parameral spine curved upwards, but straight parameral spine in A. obidensis; aedeagus with the apex of the lateral apodeme dorsoventrally flattened, but apex not flattened in A. obidensis.
Additional comments. Angiopolybia paraensis was described by Spinola (1851) and his type specimen (or specimens) was deposited in the Museo Regionale di Scienze Naturali di Torino (MRSN, Torino, Italy). Richards (1978) did not find any A. paraensis type specimen during his study about the social wasps of the Americas, and we did not receive any answer from the Museum about the type specimen. Additionally, on the online page Checklist of Epiponini wasps (http://iunh2.sci.ibaraki. ac.jp/wasp/Epiponini/epiponini.htm; consulted in 2021) produced by Dr. James M. Carpenter, the presence of the type specimen in the Museum's collection is also uncertain. Based on this, the redescription of the species was carried out using a specimen identified by Ducke in 1909 and from the same locality of the type specimen (Brazil, Pará, 26.9.1901, Ducke / Polybia paraensis Spin. ♀, det. Ducke 1909/ Brazil., Mus.Goeldi., 1910). The information about the male specimen described is: BRA, Roraima, Amajarí, Serra do Tepequém, SESC Tepequém. 1-15.iii.2016 / Malaise grande, J.A. Rafael, F.F. Xavier Filho col.

Type specimen. Without information.
Additional material examined. We examined 79 females and six males for A. paraensis; see supplementary material S1.

Key to Angiopolybia species
The following key is a revised and adapted version with few modifications of the keys provided by Richards (1978) and Andena et al. (2007).

Phylogeny and molecular species delimitation
Our concatenated data matrix of 934 aligned base pairs (bp) contained 166 variable sites, composed of 415 bp of Cox1 and 519 bp of 16S. Our morphological dataset was composed of 17 binary characters, six multi-state characters, and two contingent characters. The characters 1-18 and 19-25 are from female and male adults, respectively, as listed below and shown in Fig. 8 (coded in Table S4). The characters 2, 4, 12, 14 were modified from the study of Andena et al. (2007), and the characters 7-10, 13, 15-17, 19-25 are proposed in this study.

Morphological characters
duced posterolateral carina (character 11, state 1), paramere spine with base of the ventral margin non-angulate (character 20, state 1), digitus with rounded anteroventral lobe (character 22, state 1) (Fig. S1). The genus has two well-defined clades. The first clade is formed by A. pallens and A. zischkai (with anterior wing 7-9.5 mm long) (SH-aLRT: 95.3%, UFboot: 97%, and MP Bootstrap: 96%). These two species were recovered as sister groups on MP by the synapomorphy of the axillary fossa with anterior margin directed to the posterior region (character 13, state 1). The second clade is formed by A. paraensis and A. obidensis (with anterior wing 12-16 mm long) (SH-aLRT: 99.8%, UFboot: 100%, and MP Bootstrap: 93%. These two species were recovered as sister groups on MP by the following synapomorphies: lateral ocelli on a declivity at the vertex (character 1, state 2), eyes composed with very small bristles (character 2, state 1), angulate gena with lower region broader (character 4, state 2), apical portion of the aedeagus with the median region of the reduced denticulation (character 24, state 1), apical portion of the aedeagus curved to the venter (character 25, state 1) (Fig. S1).  The four molecular species delimitation hypotheses (ABGD, ASAP, bPTP, and mPTP) ranged from three to six species (Fig. 9). The Cox1 and 16S showed different results such as Cox1 delimiting between 3 (ABGD and mPTP) and 6 lineages (ASAP), the 16S between 1 (mPTP) and 4 lineages (ASAP), and the concatenated fragments (Cox1 + 16S) between 2 (mPTP) and 5 lineages (ASAP and bPTP). Angiopolybia obidensis and A. paraensis showed the lowest interspecific genetic distance, 5.37% for Cox1 and 1.89% for 16S. The highest interspecific genetic distance is between A. paraensis dark brown and A. pallens from Ecuador with 18.14% for Cox1. Between A. zischkai and A. pallens the lowest genetic distance is 10.15% for Cox1. The highest intraspecific genetic distance was found for A. pallens, with 9.49% among specimens from the Brazilian states Rondônia and Bahia (all values of genetic distance of Cox1 and 16S are listed in Table S5). Unfortunately, the intraspecific genetic distance of A. zischkai and A. obidensis was not analyzed because only one specimen of both species had the DNA extracted. These species, despite showing slight variation in the color of some populations, are morphologically corroborated considering the validity of their taxonomic status.

Discussion
Our ML and MP inferences with concatenated data corroborate the monophyly of Angiopolybia (Wenzel and Carpenter 1994;Andena et al. 2007;Menezes et al. 2020;Noll et al. 2021). The phylogenetic relationships recovered in our study for Angiopolybia species were similar to those found in the phylogenetic hypothesis proposed by Andena et al. (2007) based only on morphological data. Richards (1978) also described the morphological approximation among Angiopolybia species, despite not using a phylogenetic approach. Based on the morphological and molecular analysis, the variants of A. paraensis described as Polybia ruficornis Ducke, 1904 (A. paraensis morph ruficornis denominated by Richards, 1978) and Stelopolybia paraensis var. (or subspecies) obscurior Bequaert, 1944 (A. paraensis morph obscurior denominated by Richards, 1978), and variant A. pallens dark morph described by Richards, 1978 are not distinct, and we do not recommend the use of the term for species, as also suggested by Andena et al. (2007).
A high genetic diversity of A. pallens was also found by Carvalho et al. (2021) using three mitochondrial markers. However, the genetic relationship found by the authors between the populations from Acre and Rondônia with those from Bahia, Pernambuco, and Rio de Janeiro was smaller than the relationship with other populations from the Amazon Forest. The high genetic distance detected in A. pallens is possibly due to its wide geographical distribution, timing of colonization, and climatic stability in remote areas in both Amazon and Atlantic forests (for details see Carvalho et al., 2021).
Based on our morphological and molecular results, the color variation found in A. pallens, showing a gradient from yellow to black (Fig. 7a-g), represent only intraspecific variation for this species. In social wasps, phenotypic variants related to color can be derived from genetic differences, abiotic factors, aposematism, and Müllerian mimicry (Perrard et al. 2014;Badejo et al. 2020). A study investigating the color variation versus altitude for Agelaia pallipes (Olivier, 1792) verified an interesting melanization process of the tegument related to the elevation of the Andean altitude, suggesting that the darkest color in this species is associated to thermoregulation and photoprotection (Souza et a. 2020). Further research is necessary to understand the possible causes of color variations in A. pallens, since both yellow and dark forms were found at low and high altitudes (see Fig.  S2 and Supplementary material S1). For instance, species with darker coloration of velvet ants (Mutillidae) are frequent in habitats with higher humidity, more vegetation, and UV-B radiation (Lopez et al. 2021). For A. paraensis, the color variants found are well-defined because they occurred in geographic-close regions and often in the same locality (Fig. 10d).
The different results by the four molecular species delimitation methods used for Angiopolybia species can be interpreted by the limitations to the dataset and approach used here as well as a possible overlap between intraspecific and interspecific limits, as verified in the cicada Tettigettalna (Nunes et al. 2014). The missing information considering both taxonomic sampling and molecular dataset may have reduced the statistical power of our analyses. Additionally, because we used only mitochondrial markers (they are supposed to be linked, non-recombining locus with maternal inheritance, and likely evolved under similar constraints) and a category of species delimitation methods threshold-based that propose species hypotheses from a single-locus, our molecular species delimitation results should be viewed with some caution. Following the general lineage concept (de Queiroz 2007) in the search for separately evolving lineages that represent the species of Angiopolybia, we opted to be conservative considering our molecular species delimitation results, as recommended by Carstens et al. (2013), since here we take advantage of using both morphology-based taxonomy and molecular-based approaches. Although A. paraensis dark brown can be considered as such an intraspecific evolutionary lineage, additional molecular and morphological (e.g., male genitalia) data will be necessary to propose as a new taxon. Our results are useful for future biogeographical studies and provide an integrative framework that could be applied to the taxonomy of other Neotropical social wasps.

Competing interests
The authors have declared that no competing interests exist.

Acknowledgements
We are thankful to the curators and assistants of the zoological col-