All analyses recovered Augochlora as monophyletic with strong support values (Fig. 1) and high posterior probabilities (PP > 95, Fig. 2, Supplementary file 4: Phylogenetic trees, Fig. 1). According to implied weighting and partitioned Bayesian analyses Augochlora is sister group to Augochlorella. However, unpartitioned Bayesian analysis found Augochlorella as the sister group to the remaining genera while Augochlora as related to Ceratalictus + Pereirapis (Supplementary file 4: Phylogenetic trees, Fig. 2) and parsimony under equal weights recovered a polytomy at this node (Supplementary file 4: Phylogenetic trees, Fig. 3). For simplification purposes, support values and posterior probabilities mentioned henceforward are based on results of the implied weighting parsimony and partitioned Bayesian analyses.

Both valid extant subgenera of Augochlora were recovered as monophyletic. The subgenus Oxystoglossella has 73 bootstrap and 80 jackknife support values (Fig. 1) and PP value equal to 0.95 (Fig. 2). Its synapomorphies are the pale yellowish hind basitarsus on males and the setae on the outer lobe of ventral gonostylus longer than inner lobe length (Fig. 1). The relationships among species are stable among analyses and two main clades are found, herein the A. iphigenia and A. morrae species groups. Each species group has three synapomorphies, with the basal elevation of the labrum slightly produced in A. iphigenia group and orbicular in A. morrae group. The A. iphigenia group (PP = 0.96) is also defined by the broad mandible and not expanded preoccipital carina, and A. morrae group (PP = 0.97) by slightly pointed hypostomal carina and long S7 pre-apodemal projection.

Augochlora s. str. have 72 bootstrap and 82 jackknife support values (Fig. 1) and PP value equal to 0.73 (Fig. 2). Its synapomorphies are the thin apical black band on clypeus and the metapostnotum about 1.5x longer than metanotum, and the black basal area of the labrum (only at fast optimization). There are five major clades within Augochlora s. str. recognized here as species groups. Despite the low support values in the parsimony analysis, relationships among the species groups are relatively well supported in the Bayesian analysis (PP > 0.7). All species groups defined herein were stable among analyses and have posterior probabilities higher than 0.95, with the exception of the A. pura species group. Augochlora (A.) intermedia sp. nov. has an isolated placement and is sister group to the remaining groups according to three out of four analyses (Fig. 1–2). The remaining Augochlora s. str. are grouped by three non-homoplastic synapomorphies: preapical tooth of mandible sharp and produced near apex of mandible, pseudo-pygidial area without appressed scale-like setae, and T1 with a relatively large impunctate area near apical dark band. On the other hand, the unpartitioned Bayesian analysis recovered the A. hestia species group as the sister group with the remaining Augochlora s. str. (Supplementary file 4: Phylogenetic trees, Fig. 2). This group has a strong support (PP = 1) and several synapomorphies, including the orbicular elevation and yellowish basal area of labrum, features shared with some Oxystoglossella. Also, females in this group have a flattened mesofemur, with a characteristic depression near the brush posteriorly.

The remaining Augochlora s. str. species are grouped by the presence of a long dark area between the premarginal setae apex and the T3 apex, the acute shape of apical spine on outer surface of hind tibia, and by the S2 apex acute medially on males (Fig. 1). The presence of a median longitudinal impression on the scutellum is a synapomorphy of the A. daphnis species group (PP = 0.98), being valuable to diagnose the group. Another synapomorphy for this group is the angulation between the anterior and posterior surfaces of the hind coxae, sometimes forming a slight longitudinal carina, a condition also found in A. (A.) intermedia sp. nov. The A. repandirostris species group has strong support (PP = 1) and several synapomorphies, some of them useful to readily distinguish its species, such as the projected anterior portion of the clypeus, the less projected epistomal angle, and the depressed face around the antennal sockets. The A. repanditrostris and A. pura species groups are defined as a clade by the absence of tiny setae among long setae on the outer surface of hind tibia and the inflexed apex of T1, both features observed in males. The A. pura group (PP = 0.62) is singly supported by homoplasies and relationships among some clades are not clearly identifiable through morphology. While the consensus trees of parsimony analyses resulted in a polytomy at the base (Fig. 1 and Supplementary file 4: Phylogenetic trees, Fig. 3), the Bayesian analysis with partitioned data indicated A. (A.) nausicaa as the sister group with remaining species and A. (A.) pura and A. (A.) dorsualis as related to a clade with A. (A.) cydippe.

Herein we analyzed a morphological matrix for Augochlora under two different approaches – parsimony criterion and Bayesian inference. The genus and both extant subgenera were recovered as monophyletic, with relatively high support values on all analyses. Within the genus, both approaches provided very similar topologies, allowing us to consistently identify major lineages. Partitioning using characters’ level of homoplasy resulted in trees more alike those generated by implied weighting parsimony, which is expected since both approaches do not treat characters as equally influential for topology estimation. This kind of approach yielded trees with outgroup relationships identical to those recovered by molecular and total-evidence analyses (Gonçalves 2016; Gonçalves et al. in press). The use of parsimony is widespread for morphological datasets, probably as a reflex of tradition and simplicity. On the other hand, the use of model-based approaches for phylogeny estimation of discrete morphological data is increasing and revealing many advantages over other methods (see Wright and Hillis 2014’ O’Reilly et al. 2016). We understand that the concomitant use of parsimony and Bayesian approaches helps to better explore available information.

One of our main goals was to evaluate the valid subgeneric classification (sensu Engel 2000, Michener 2007; Moure 2007). As both extant subgenera were recovered as monophyletic, their use should be maintained for stability. To achieve this, a species placed in Oxystoglossella in the current classification (Moure and Hurd 1987; Moure 2007), A. bractealis should be transferred to Augochlora s. str. This species superficially resembles Oxystoglossella, and also have some labral modifications shared with that subgenus, but it belongs to the A. hestia species group in Augochlora s. str. (synapomorphies supporting a close relationship between A. (A.) bractealis and A. (A.) hestia are shown in Fig. 1). Our results support the use of Aethechlora as a separate subgenus for A. iphigenia species group, but this can unecessarely inflate the taxonomy of the group and besides the original description by Moure and Hurd (1987) no other author considered the validity of this subgenus. We also favor maintaining Augochlora s. str. as a subgenus with no further modifications, since the use of Mycterochlora as a separate subgenus for the A. repandirostris species group would unnecessarily require raising three new subgenera. The unique A. (A.) intermedia sp. nov. is remarkable in having mandibles and pseudo-pygidial area modifications characteristic of Oxystoglossella while its genital capsule is typical from Augochlora s. str. Despite the monophyly of the subgenera, we found that characters frequently used to separate them (Eickwort 1969; Engel 2000) are variable along the phylogeny. The single exception is the length of the setae on the outer lobe of gonostylus – the long setae are a synapomorphy of Oxystoglossella, while Augochlora s. str. have short setae, as in the outgroup.

The large and produced preapical tooth was considered as a characteristic feature of Augochlora s. str. females (Eickwort 1969; Engel 2000), and besides A. (A.) intermedia sp. nov., the remaining species of the subgenus have this modification. Although mandibular teeth can bear some degree of wear depending on the specimens, it is clear that some species have a broadly rounded preapical tooth, less detached from the upper margin of the rutellum. The mandibles also have distinct modifications in other clades. In the A. iphigenia species group they are elongated with a broad apical tooth, a condition taken to the extreme in some macrocephalic forms, as in A. (O.) matucanensis and A. (O.) phoenicis. In some species of the A. pura group, there is a strong constriction before the distal part of aductor ridge in frontal view, as was noted in A. (A.) atlantica, A. (A.) francisca and A. (A.) mulleri, also occurring in A. (A.) genalis Lepeco & Gonçalves (Lepeco and Gonçalves 2020a: Fig. 1). The shape of the elevation on the basal area of the labrum was also considered diagnostic of the subgenera, with the transverse elevation being typical for Augochlora s. str. and an orbicular shape for Oxystoglossela (Engel 2000). However, the shape is quite variable among species. The orbicular form is found only in the A. morrae group from Oxystoglossella and also on the A. hestia group. A. (A.) iphigenia also has a transverse elevation as interpreted by Gonçalves (2019), and its allied species have a slight elevation, without a well-defined shape, as discussed for A. (O.) modica and A. (O.) tenax by Lepeco and Gonçalves (2020b).

We noted that the pseudo-pygidial area on the fifth tergum is a key element in the systematics of Augochlora, but it has been described in different ways by previous authors. Eickwort (1969) considered that species in Augochlora s. str. bear scale-like setae in the pseudo-pygidial area of T5, while Oxystoglossella lack these setae. This character was not mentioned by Engel (2000), but posteriorly Dalmazzo and Roig-Alsina (2011) considered that species of Oxystoglossella bear setae-like spicules, while Augochlora s. str. bear globose spicules (see Figures 11 and 12 of the latter work). Lepeco and Gonçalves (2020a) suggested that the structures found in Augochlora s. str. are simply the colliculate integument exposed due to the absence of scale-like setae found in Oxystoglossella. All species of Oxystoglossella and the outgroup bear appressed setae on the pseudo-pygidial area, somewhat resembling scales, while most Augochlora s. str. lack these setae and have a more sclerotinized irregular surface on each side of the medial cleft of T5. This irregular surface is formed by many tiny protuberances, clearly shown by Dalmazzo and Roig-Alsina (2011: Fig. 12), that are not similar with setae or scales. We maintain the interpretation of Lepeco and Gonçalves (2020a), adding the indication that the setae found in the medial area of T5 in Oxystoglossella are appressed, to avoid confusion with the terms coined by Eickwort (1969) and Dalmazzo and Roig-Alsina (2011). The presently described A. (A.) intermedia sp. nov. is the single Augochlora s. str. to bear appressed scale-like setae on the pseudo-pygidial area, although the coverage is sparser than observed in Oxystoglossella, maybe representing an intermediate state.

Augochlora is a widespread genus in the Western Hemisphere, but much of its diversity and distribution patterns are largely unknown. In the case of the revised fauna from Southern South America (Lepeco and Gonçalves 2020a), it is clear that species exhibit different ranges and habitat tolerances, with some of them occurring in areas strictly covered by rainforests and others also reaching regions of open vegetation. An important observation derived here is that the main lineages, the species groups, have a widespread distribution on western hemisphere. Our taxon sampling is inadequate for a formal biogeographic reconstruction, as we did not sample regions equally within species groups, leaving many distributional gaps in our data. In addition, the diversity of some regions, such as Mesoamerica and Amazonia, is not fully known, leaving an obstacle for any phylogenetic and biogeographic exploration. Therefore, in light of our knowledge on the subject, we bring some insights also considering species not represented in our phylogeny.

Oxystoglossella is less speciose in comparison to Augochlora s. str., allowing us to identify some distribution patterns based on the studied species. The A. iphigenia species group is restricted to South America. Within this group, A. (O.) modica and A. (O.) tenax are recovered as sister species in our results, being typical from the Caatinga in Northeastern Brazil, a region characterized by dry conditions and high temperatures (Lepeco and Gonçalves 2020b). Augochlora (O.) mendax Lepeco and Gonçalves, 2020 is very similar to A. (O.) modica, and is found throughout the Cerrado formation, in central Brazil. A fourth species is known to us only from the Serra do Cipó locality, in Minas Gerais, Brazil, characterized by dry conditions. It seems clear that these four species presumably form a monophyletic group well-adapted to Brazilian open vegetational formations, being the single group bearing this characteristic within the genus. On the other hand, the clade composed of A. (O.) iphigenia, A. (O.) phoenicis and A. (O.) matucanensis is more heterogeneous. A. (O.) iphigenia is abundant from central Brazil to the Argentine Pampas, while A. (O.) phoenicis is found only within Western Amazonia, and A. (O.) matucanensis is the single species in the subgenus found in the Pacific side of the Andean mountain range in Peru. The A. morrae species group is mostly composed of commonly collected species, distributed throughout the almost entire range of the subgenus, with A. (O.) vincentana occurring at Caribbean islands (Moure 2012). Another species, not studied here, was described for Haiti: A. (O.) haitiensis (Vachal, 1911). Finally, a few species of Oxystoglossella occur in Central America, and at least three species occur in Mexico, with A. (O.) aurifera Cockerell, 1897 also recorded for the USA (Ascher and Pickering 2021).

Within Augochlora s. str., the A. daphnis and A. pura species groups are the more widespread, with the later also being the more speciose and morphologically diverse. Representatives of both groups are found in the USA, with A. (A.) pura reaching Canada and A. (A.) nigrocyanea Cockerell, 1897, a species undoubtedly belonging to the A. daphnis group, reaching southern USA (Sandhouse 1937; Moure 2012). Our results indicate that the occupation of the Nearctic region occurred within few derived taxa, disfavoring a scenario where Augochlora originated in North America. Interestingly, both species groups are found in the southernmost locality where Augochlora were ever collected – the Neuquén Province in Argentina – since A. (A.) australis Lepeco and Gonçalves, 2020 and A. (A.) hirsuta Lepeco and Gonçalves, 2020 are probably related to A. pura and A. daphnis species groups respectively. All Augochlora s. str. known to us from the Caribbean islands seem to be related to the A. daphnis species group, and, according to our results, the occupation of Caribbean islands occurred at least three times within derived taxa of the genus. An early diversification of Augochlora in the Caribbean islands cannot be discarded if we consider that E. leptoloba is the sister group with all other Augochlora (Gonçalves et al. in press). From A. daphnis group came the only species recorded on Fernando de Noronha archipelago (545 km from Brazilian coast), Augochlora laevipyga is related to Augochlora esox.

The A. repandirostris species group is more abundant and diverse in the Amazon rainforest, but can also be found in the Panamanian Forest (e.g., A. (A.) isthmii Schwarz, 1934), and in the Atlantic Forest, (A. (A.) helena Lepeco and Gonçalves, 2020 and an undescribed species from Espírito Santo). The A. hestia species group has a similar distribution with two species from Amazonian Forest and one species widespread in the Atlantic Forest known to date. Species from both groups seems to be restricted to rainforests, with no representative being ever recorded in the open formations of South America. They also were not collected in the Caribbean.

The western coast of Peru appears to be an important region for the biogeography of the genus, with two species occurring there – A. (A.) intermedia sp. nov. and A. (O.) matucanensis. Both species are morphologically unique, and the former is the single representative of a lineage sister to all other Augochlora s. str. Nevertheless, the diversity of Augochlora in Peru, western to the Andean mountain range, seems to be low. The original label of A. (O.) matucanensis indicates that these bees were collected at Andean foothills, where vegetational conditions may diverge from that explored by most of Augochlora species Another case to be addressed is A. (A.) notialis (Vachal, 1904), known only from the male holotype, labeled from Chile. Unfortunately, we did not examine the holotype and, according to Moure and Hurd (1987), it was probably mislabeled.

In summary we understand that the first evolutionary and biogeographic events of extant Augochlora lineages took place in South America. This seems to be evident by the diversity of Oxystoglossella, as a few groups with distinctive morphologies are restricted to South America. In the case of Augochlora s. str., two of its early branching lineages are endemic to South America (i.e., A. intermedia sp. nov. and A. hestia species group). Nevertheless, a refined elucidation of the biogeographic history of Augochlora can only be achieved with better understanding of its taxonomy and phylogeny.

Despite direct observation, the social behavior can be also positively inferred from intraspecific polymorphisms. Augochlora (O.) iphigenia, A. (O.) matucanensis and A. (O.) phoenicis are remarkable for the existence of females with distinct head enlargement, along with expansion of the apical portion of the mandible and projection of the hypostomal carina. These characteristics were also documented in A. (O.) empusa Engel, Hinojosa-Díaz and Bennett, 2012, which is probably a junior synonym of A. phoenicis. Morphological polymorphisms of this sort are often linked to social interactions within the nest, as the large foundress females of primitively eusocial bees physically subjugate their subordinate daughters (Pabalan et al. 2000; Packer et al. 2003). Allometric growth of head structures in females is also present in species of the A. daphnis group (including the facultatively eusocial A. daphnis) and in A. (A.) genalis Lepeco and Gonçalves, 2020, a putative member of the A. pura species group (Lepeco and Gonçalves 2018; Lepeco and Gonçalves 2020a). It is unclear whether such modifications are invariably related to eusociality, even so, conspicuous cephalic polymorphism seems to have evolved at least three times independently in the genus.

Michener (1990) pointed out that solitary behavior in Augochlora s. str. represented a reversal from eusociality, since Oxystoglossella and the related genera are primitively eusocial (Eickwort 1969; see also Danforth and Eickwort 1997). A. (A.) pura is a well-known solitary species, and no signals of eusociality were found in A. (A.) hallinani Michener, 1954; A. (A.) sidaefoliae Cockerell, 1913; and A. (A.) smaragdina Friese, 1916 (Stockhammer 1966; Eickwort and Eickwort 1973). It is difficult to conclude whether these species are indeed solitary, due to the low number of nests observed. Given the presence of eusociality in A. (A.) isthmii, a member of the repandirostris species group, and in A. (A.) daphnis and A. (A.) phoemonoe (Dalmazzo and Roig-Alsina 2012, 2015, 2018a, 2018b), our results point to the presence of primitively eusocial species in at least three of the five Augochlora s. str. species groups, indicating that the exclusive solitary behavior of A. (A.) pura can be a reversion case.

Eusociality has been lost many times in eusocial Halictinae lineages (Danforth 2002; Gibbs 2012). Unlike in fixed-caste eusocial taxa (e.g., Apini, Meliponini), reversal to solitary behavior does not represent a large evolutionary obstacle for the Halictinae. Since all totipotent-caste eusocial bees pass through a solitary phase during the year (i.e., nest construction and provisioning before the first brood emerging) shift to solitary behavior would be acquired by simply changing from a multivoltine to univoltine cycle, where only reproductive brood is raised (Gibbs 2012; see also Almeida and Porto 2014). In fact, eusociality is facultative in many Halictinae species, being determined at the transcriptional level and influenced by climatic conditions, with females behaving as solitary in colder regions (Sakagami and Munakata 1972; Eickwort et al. 1996; Kapheim et al. 2020). Another important factor is resource limitation, species that nest in wood often have to cope with substrates with different sizes, shapes and diggability, which probably limit colony expansion (Dalmazzo and Roig-Alsina 2015).

Among Augochlora, facultative behavior can be suggested for A. (A.) daphnis, since individual nests were found both with a single female or with many females dividing activities (Dalmazzo and Roig-Alsina 2012). This is also the case of A. (A.) isthmii (Wcislo et al. 2003) and probably for the remaining Augochlora s. str.. Other kinds of social organization, such as communal nesting, may be common within the genus, as it is the case for Halictinae (Michener 2007). We suggest that eusociality is a plastic characteristic in the genus, perhaps with multiple gains and losses along its evolution. Additionally, the eusocial behavior in Augochlora is probably facultative as a rule. Nevertheless, the evolutionary history of eusocial behavior will be only completely understood with nest observations and detailed records.

Modified mandibles and absence of appressed scale-like setae on pseudo-pygidial area are found in primarily wood-nesting augochlorines. The studied nesting species of Megalopta Smith and allied genera (Megaloptosyne Engel and Xenochlora Engel, Brooks and Yanega) nest in decaying wood, and their females have strong mandibles with supplementary teeth on the inner surface (Michener and Fraser 1978; Santos et al. 2010) and colliculate pseudo-pygidial area, without appressed setae (see Engel 2000: Fig. 59). The same T5 modification and developed preapical tooth can be found in some species of Neocorynura Schrottky which nest in wood (Eickwort 1969; Brosi et al. 2006), but in this case a phylogeny to trace the origin of these traits is lacking. All species of Augochlora s. str. studied so far nest out of the soil (Table 1), and, using the mentioned morphological clues, we suggest that this behavior evolved in the common ancestor of the A. hestia, A. daphnis, A. repandirostris and A. pura species groups. Although the new species, A. intermedia sp. nov., is more related to Augochlora s. str. species, it has mandibles typical of soil-nesting species and sparse scale-like setae on the pseudo-pygidial area. Alternatively, the new species may be a facultative wood-nester, as is the case of A. (A.) esox, recorded nesting on bromeliad humus (Zillikens et al. 2001) but also found nesting in wood (G. A. R. Melo, personal communication). Nevertheless, Augochlora s. str. may not be strictly dependent on a single type of nesting substrate, a conclusion supported also by laboratory rearing of A. (A.) pura in soil (Barrows 1973).

In regard to nest architecture, bees exhibit a certain level of plasticity, adapting the organization of structures according to substrate conditions (Eickwort and Sakagami 1979). Nests with clusters of cells supported by earthen pillars within a hollow cavity are the commonest among Augochlorini, probably representing a plesiomorphy for the Augochlora group of genera (Danforth and Eickwort 1997; Engel 2000; Wcislo et al. 2003; Coelho 2004). These clusters facilitate the isolation of cells from the surrounding soil by means of a drainage cavity, protecting the provisions and brood from excessive moisture (Sakagami and Michener 1962; Eickwort and Sakagami 1979; Wcislo et al. 2003). Clusters surrounded by cavities are present in nests of A. (A.) daphnis within soft wood, but were not found in trunks with harder wood (Dalmazzo and Roig-Alsina 2012). In the other few Augochlora nests studied, clusters are either present or not, showing no consistent phylogenetic pattern. On the other hand, all Oxystoglossella studied so far do not construct cells in clusters (Eickwort and Sakagami 1979), suggesting that their absence is not necessarily linked to wood-nesting. Nevertheless, it is not known whether the absence of clusters and/or hollow cavities surrounding cells in nests of Augochlora s. str. is related to the better drainage proportioned by wood, as suggested by Wcislo et al. (2003), or due to heterogeneity of wood material in relation to soil, where conditions allow females to express their full construction capabilities (Dalmazzo and Roig-Alsina 2012). In the latter case, environmental conditions have a major role in defining architecture in relation to phylogenetic relationships.

Decaying wood has the disadvantage of being less abundant in comparison to suitable soil, but, on the other hand, it is more difficult to become soaked during periods of rainfall and retains humidity in dry periods (Stockhammer 1966). Regarding social behavior, nesting above ground did not limit Augochlora s. str. to the solitary lifestyle (Dalmazzo and Roig-Alsina 2012, 2015, 2018a, 2018b). Examining other augochlorine wood nesters, Megalopta and allied genera can be primitively eusocial (Kapheim et al. 2020), but for Neocoynura this behavior was not yet published (Brosi et al. 2006) and there are no morphological clues such as female polymorphisms. Michener (1990) suggested that the scarcity of suitable plant material, in contrast to soil, triggered the reversion to solitary behavior, as natural enemies in forest environments were less successful and, therefore, disadvantages of eusociality outweighed the advantages. This argument, if valid, is restricted to temperate species A. (A.) pura so far.

As is common for bees, Augochlora are parasitized by conopid flies and mutillids (Michener and Lange 1958). The single putative cleptoparasite bees to attack nests of the genus are Temnosoma Smith: females of an unidentified species were already seen flying around and found within nests of A. (A.) esox (Silveira et al. 2002). Eickwort (1969) suggested that Temnosoma were cleptoparasites of eusocial species, given their distinctively coarse integuments. The association of Temnosoma with Augochlora is supported by the distribution of both genera – these are the only augochlorine genera occurring simultaneously at USA, Chile and the antilles, besides most of the Neotropical region (Eickwort 1969; Michener 2007; Moure 2012). If Temnosoma did not parasitise other groups, this association may represent an adaptation to host-seeking in an unusual substrate, i.e., decaying wood. A similar trend is found on the cleptoparasitic species of the wood nester Megalopta (Michener 2007). Considereing this hypothesis, the utilization of decaying wood for nesting may have facilitated the origination of cleptoparasitic lineages within the tribe, with less competitors, since other soil nesting genera (e.g., Augochlorella) are usually attacked by the widespread genus Sphecodes Latreille (Ordway 1965).