The specimens examined belong to the following institutions: American Museum of Natural History, New York, USA (AMNH); Instituto Nacional de Pesquisas da Amazônia, Manaus, Brazil (INPA); Montana Entomology Collection, Bozeman, USA (MTEC); Museu de Zoologia da Universidade de São Paulo, São Paulo, Brazil (MZUSP); Museum für Naturkunde, Berlin, Germany (MfN); National Museum of Natural History, Smithsonian Institute, Washington, D.C., USA (USNM); Natural History Museum, United Kingdom, London, England (NHMUK); Senckenberg Naturhistorische Sammlungen Dresden, Dresden, Germany (MTD); and Staatliches Museum für Naturkunde Stuttgart, Stuttgart, Germany (SNMS).

Terminology follows Cumming and Wood (2017) for general morphology. The abbreviation list used in this paper follows: I, II, III: fore-, mid-, hind legs; CI: consistency index; Cx: coxa; F: femur; L: length; MSSC: male secondary sexual character; RI: retention index; T: tibia; t_1–5: tarsomeres 1 to 5. Photos were taken using an AxioCam MRc 5 camera attached to a Zeiss Discovery V20 stereomicroscope and stacked with ZEISS AxioVs40 v. 4.8.2.0 software and assembled in Helicon Focus 6. For examination under microscope, the terminalia were treated with lactic acid following the procedure described by Cumming (1992), and after examination transferred to a microvial with glycerin attached to the same pin of the dissected specimen. The final images were worked on Adobe Illustrator CS6 for the separate pictures and Adobe Photoshop CS6 for the plates.

Parsimony analyses were carried out in the TNT version 1.6 program (Goloboff 1993; 1995; Goloboff et al. 2008; Farris 2008; Goloboff and Morales 2023) through heuristic searches using equal weighting (EW) and differential weighting of characters by implied weighting (IW). The following parameters were used for traditional search: maximum of 20000 trees in memory; random seed = 0; repls. = 1000; trees saved per replication = 20, the trees collapsed after analysis. Bremer support was established using the command prompt “sub 1 hold 6000; bb = tbr fillonly; unique*”, changing the value of suboptimal trees. The ordering of the transformation series was carried out by rooting with outgroups (Nixon and Carpenter 1993).

The dataset transposed into the matrix comprised morphological characters of adult males and females (Table S1), meticulously curated using Morphobank (O’Leary and Kaufman 2007) version 3.0. New characters were coded following appointments of Fitzhugh (2006) and Sereno (2007) (see Results: Character List). The ingroup includes all 12 known nominal species of the genus Achradocera (Quevedo et al. 2024): A. angustifacies Becker; A. apicalis (Aldrich); A. arcuata (Van Duzee); A. balin Quevedo, Capellari and Lamas; A. barbata (Loew); A. excavata (Van Duzee); A. femoralis Becker; A. gimli Quevedo, Capellari and Lamas; A. insignis Parent; A. longiseta Parent; A. meridionalis Becker; and A. tuberculata (Van Duzee). Species from three different genera of the same subfamily (Diaphorinae) as Achradocera composed the outgroup: Chrysotus Meigen [4 spp.]; Diaphorus Meigen [1 sp.] (root); and Lyroneurus Loew [1 sp.]. The program WinClada ASADO 1.61 (Nixon 1999–2002) was used to analyze the trees. Brackets are used to indicate the numbers of characters from the character list, and points are used after the numbers to indicate the character status (e.g., [2.1] indicates character 2, state 1).

The analysis incorporated data from a total of 204 localities (Table S2), sourced from the labels of the specimens under study. One record was used from iNaturalist [photographic record: https://www.inaturalist.org/observations/108430413] and another was provided by personal communication [Mexico. 1m#, “Ciudad de México, Universidad Nacional Autónoma de México, Cantera Oriente, La grieta, 19.31639, -99.17179, 06.iii.2024, L. R. P. Gomes”]. Achradocera specimens from Hawaii, Tonga and French Polynesia were probably accidentally introduced in those islands (Bickel 2000; Evenhuis 2012) and the data was not plotted to avoid noise in the analysis. All collected data were plotted on a New World shapefile and edited in QGIS 3.4 (QGIS Development Team 2020).

Event-based analysis was conducted employing the Geographically explicit Event Model (GEM) method, facilitated through the EVS software package (Arias 2017). EVS configuration adhered to the author’s guidelines, employing a raster grid with 1x1 degree pixels and a fill value of 1. Each of the four events (vicariance, sympatry, founder event, and point sympatry) carried an equal cost of 1. To penalize ancestral distributions and deter extensive ancestors, a value of Z = 10 was employed, where Z denotes the size of ancestral ranges. The analysis employed a flipping algorithm with 10 independent runs, each comprising 10,000 replicates.

The EW analysis yielded two Most Parsimonious Trees (MPTs) with a length of 117 steps, CI of 0.62 and RI of 0.71 (Fig. 3, T1 and T2). The strict consensus tree exhibits a length of 118 steps, CI of 0.61 and RI of 0.71 (Fig. 3, T0). All searches utilizing IW resulted in only one topology (Fig. 3, T1), regardless of the value employed k. This suggests that any minimal weighting against homoplasy favors T1 as the most parsimonious topology. Therefore, it serves as the primary evolutionary hypothesis for this study. However, T2 will also be considered in the discussions.

Achradocera is robustly supported as a monophyletic group, characterized by six synapomorphies: enlarged postpedicel with a broad base constricted into a long narrow tip [0.1]; postpedicel length slightly longer than arista [1.2]; flattened ventral postocular setae [3.1]; grey pruinosity restricted to the area close to the eye margin [6.1]; lateral lobe of epandrium appressed on the margin of the epandrium [42.2]; and phallus with conspicuous lump-like grooves on dorsal surface [49.1]. The high Bremer’s support (7) further corroborates the strength of this clade.

Within Achradocera, it is also possible to delineate smaller lineages of species. The femoralis- and barbata-groups are easily recognizable due to their morphological traits. The barbata-group comprises A. barbata and A. arcuata, both displaying numerous synapomorphies on the tarsomeres of the male foreleg [19.1 and 20.1] and hypopygium [37.3, 39.1, 40.1, 45.1, 49.2, 50.1, and 51.2] (see Figs 1D and 2A–C). Similarly, the femoralis-group―consisting of A. excavata, A. femoralis, A. meridionalis and A. tuberculata―is well supported, sharing many synapomorphies on the male legs [18.2, 25.1, 26.1, 27.1, 28.1, 29.1, and 30.1] (Fig. 1B, E, F, H, I). The femoralis-group is sister of A. angustifacies and both share the apomorphic state of character 32.1 (IIIt_3–4 concave) as a synapomorphy with all the other species (see Fig. 1E). However, species of the femoralis-group accumulated many other synapomorphies after cladogenesis with A. angustifacies, and maintenance of the femoralis-group remains more applicable due to the ease species recognition associated with those characters that appears during the anagenesis of the group.

The last lineage that deserves further attention is composed by A. apicalis, A. gimli and A. balin. The primary challenge is that this clade lacks any exclusive synapomorphy and is solely grouped based on the infuscated pattern on TI and TII (see remarks of character 15). Moreover, Bremer’s support for this clade was low, and is preferable to avoid the use of “species group” for this assemblage of species. Ultimately, A. longiseta was recovered as the sister-group of the remaining species and, in this latter group, A. insignis is the sister of all other species. The apomorphic dense white “beard” composed by the ventral postocular setae [3.1] is distinctive of the entire genus, yet in A. longiseta this trait is not remarkably developed. Additionally, this species also exhibits rather simple legs, which justifies its position as the sister group of all the other Achradocera species.

The analysis recovered Chrysotus spectabilis as the sister group of Achradocera. However, assigning this relationship is challenging given the complexity of the genus Chrysotus, which comprises over 300 species and is likely to be polyphyletic (Capellari and Amorim 2010, 2012, 2014) and even in our current analysis, we recovered Chrysotus as paraphyletic. Hypopygial characters favored those results, nesting C. wirth and Lyroneurus adustus [41 and 46] and placing C. spectabilis as the sister-group of Achradocera [47]. Van Duzee (1924) formally split Chrysotus into smaller species-groups to enhance our understanding of the genus, but his units are a mix of shared overall similarities and not necessarily reflect true homologies between species (see examples in: Capellari 2015; Capellari and Almeida 2024). Although the taxonomic delimitation of Chrysotus is entangled with Achradocera, a clear-cut definition of Chrysotus is beyond the scope of this work and was preliminary discussed by Capellari and Amorim (2012).

The GEM method was applied for the two MPTs, with the same parameters. Both analyses yielded only one reconstruction each (Fig. 4, T1 and T2). The cost of T1 was 37, with six vicariances (Fig. 4, nodes B, C, D, F, I, and K), one point sympatry (Fig. 4, node E), three founder events (Fig. 4, nodes A, H, and J) and zero sympatries. Conversely, T2 exhibited a cost of 31, with six vicariances (Fig. 4, nodes B, C, K, F’, H’, and J’), one sympatry (Fig. 4, node E’), three founder events (Fig. 4, nodes A, D’, and G’), and zero point-sympatry. However, despite T2 displaying a lower cost, the primary hypothesis to be considered will be T1. This determination stems from the phylogenetic analysis of the taxa, where T1 was the sole topology recovered in the IW analysis and demonstrates greater robustness compared to T2.

Given the current biogeographical scenario (Fig. 5), it is evident that vicariance played a significant role in driving the majority of speciation events in Achradocera. Many species within the genus exhibit extensive distribution ranges, overlapping several areas commonly utilized as biogeographical boundaries (Morrone 2017). This phenomenon underscores both the remarkable resilience of these species across diverse biomes and their exceptional dispersion capabilities. The first vicariant event on phylogeny divides A. insignis from the ancestor of the clade C (Fig. 4). Subsequently, we can observe the fragmentation between the ancestors of the species from node D and the Nearctic clade (node K), which species were also originated by vicariance. Notably, A. arcuata is restricted to the western side, whereas A. barbata occurs on the eastern side of North America (Fig. 5).

It is noteworthy to observe the reconstruction depicted in clade D (Fig. 6). Following the separation from Nearctic species, the subsequent vicariant event delineates the ancestor of the femoralis-group (Fig. 6D: yellow squares) on the west side of the Andes, while the hypothetical ancestor of the clade (A. gimli (A. apicalisA. balin) (Fig. 6D: red squares) remains restricted to the east side of the Andes. While the clade (A. gimli (A. apicalisA. balin) exhibited only point sympatric speciation, the femoralis-group undergoes alternations between vicariances and founder events.

Examining Figure 6, we note that the subsequent vicariant event isolates the ancestor of A. angustifacies in Chile [clade F]. Following this, a founder event drives the evolution of A. femoralis in the northern region of South America [H]. Another vicariance event then fragments the ancestor of clade I, leading to the scenario where A. excavata evolves isolated on the Greater Antilles. Finally, a founder event results in the speciation of A. tuberculata, which recolonizes the eastern side of the Andes in South America [J], while A. meridionalis evolves across the western side of the Andes and the Neotropical regions of Mexico.

The GEM analysis reconstructed the ancestral distribution of Achradocera as encompassing the entire New World. However, we posit that this conclusion may have been influenced by the lack of information regarding the sister group of Achradocera, as mentioned earlier. Chrysotus, being a vast and imprecisely delimited genus, imposes a challenge to elucidate the biogeographical relationships of Achradocera at this point. As such, determining the ancestral distribution of the genus was not the primary objective of this study, given the surrounding uncertainties. Nonetheless, we can still speculate that the genus likely originated in the Neotropical region, known for its high species richness and hosting species exhibiting ancestral states of the hypopygium (Quevedo et al. 2024). In any case, a better understanding of Chrysotus is necessary for a more accurate analysis of the ancestral distribution of Achradocera.

The results of the phylogenetic analysis shed light on the evolutionary relationships within the genus Achradocera, which emerges as a monophyletic group, characterized by six morphological apomorphies pointing to a robustly defined group. The placement of C. spectabilis as the sister-group of Achradocera underscores the importance of analyzing hypopygial structures for a better understanding of the relationships between the Chrysotus groups and related genera. Certainly, the statement of the species Chrysotus spectabilis as sister-group of Achradocera should not be seen as definitive, since many other Chrysotus were not included in the analysis, even so, the hypopygial similarity of this species (Capellari and Amorim 2010) with the plesiomorphic condition of the hypopygium of Achradocera (Fig. 2D) is striking. However, the application of the same treatment to hypopygial structures may not be as useful for separating species within Achradocera—or perhaps even for the entire Diaphorinae—compared to other groups such as the longipalpus-group (Capellari 2015; Runyon and Capellari 2018), where hypopygial structures show minimal variation. This contrasts with other subfamilies of Dolichopodidae, such as Neurigoninae (Naglis 2001a, b, 2002a, b, 2003), Dolichopodinae (Soares et al. 2023; Brooks 2005), some Sympycninae lineages (Bickel 1992, 1999), and others, where hypopygial structures exhibit more pronounced differences between congeneric species.

Furthermore, our understanding of the biology and behavior of long-legged flies remains limited, particularly when contrasted with the taxonomic diversity within the group. However, the significance of their legs in courtship is apparent, as evidenced by the abundance of morphological specializations observed across various genera and species (Land 1993a, b; Lunau 1992, 1996; Zimmer 2000, 2003). In the case of Achradocera, it appears that certain convergent traits have emerged over the course of species evolution, this observation may suggest the utilization of these podomeres since their more basal lineages.

For instance, similar variations are observed in different species: A. excavata, A. meridionalis and A. tuberculata exhibit distinct setae on TI [char 17.1], while the barbata-group displays modified It_1–2 [19.1, 20.1]. Additionally, A. insignis (see Fig. 1G) independently developed an arching on TII [char 23.1], differing from the excavation observed in the femoralis-group (see Fig. 1F–H), and these characters clearly evolved independently: the excavation within the femoralis-group occurs initially with a bulge near the middle of TII, and gets more pronounced in A. tuberculata, which bears a secondary bulge more apically, forming a deep excavation. In contrast, the excavation in A. insignis is formed by an arching of the podomere, with no visible bulge, indicating the homoplasy. Achradocera balin demonstrates an arching on IIIt₂, whereas the femoralis-group and A. angustifacies display arched IIIt_3–4. Moreover, A. arcuata exhibits a modified FIII with protrusions and special setae, contrasting with the bulged FIII and special setation observed in the femoralis-group (see Fig. 1I). These observations suggest that each of these podomeres likely plays a fundamental role in courtship behavior across various species within the genus.

Due to Dolichopodidae being a highly diverse family, the focus of study on the family has traditionally been the taxonomic description of new taxa (as pointed by Lim et al., 2010–see Discussion section), which indeed is crucial, especially considering the gaps in knowledge about the fauna of naturally megadiverse regions, such as the Neotropical and Afrotropical regions (Yang et al., 2006; Grichanov and Brooks, 2017). However, some studies involving phylogeny and biogeography may represent an important tool to resolve some historically problematic groups (as advocated by Capellari and Santos, 2012), such as Chrysotus, Sympycnus Loew or various other cosmopolitan and imprecisely delimited genera.

As for the biogeographical results, although there is no known fossil of Achradocera, it is presumable that it is a fairly recent group, considering that it is a derived lineage within Chrysotus (Capellari 2013), which has Baltic fossils dated from the Eocene/Oligocene (Evenhuis 2017). Due to the locality of its most basal species (southern South America), Achradocera potentially has a Neotropical origin, later spreading to the Nearctic. According to Sanmartín and Ronquist (2004), these regions connected twice in the last 40 Myr: first with the formation of the Panamá Island Arc (15 Myr), and subsequently through the Isthmus of Panama (3.5 Myr), in a process known as the Great American Biotic Interchange. Considering that cladogenesis in clade C occurred after one of these events, it leads us to presuppose an older (h1, hypothesis 1) or more recent (h2, hypothesis 2) diversification origin of the species.

It is plausible to assume that the vicariant event observed in clade C is related to the Mexican Transition Zone (MTZ) (Morrone 2004, 2006), causing a spatial disjunction between the Nearctic clade and the other species of clade D. Linking this vicariant event associated with the MTZ to a precise time frame is challenging. Morrone (2015b) recognizes five stages in the development of the MTZ: (1) Jurassic–Cretaceous, (2) Late Cretaceous–Palaeocene, (3) Oligocene–Miocene, (4) Miocene–Pliocene, and (5) Pleistocene. Considering hypotheses h1 and h2, the vicariance of clade D may be associated with stages 3, 4, or 5. This pattern was also observed in analyses with the genus Heterostylum Macquart (Lamas et al. 2014), potentially being congruent. In the Nearctic, the vicariant event in clade K promotes segregation between the western side for A. arcuata and eastern for A. barbata. Sanmartín et al. (2001), in a biogeographical analysis involving numerous Holarctic clades for different periods of time, separate the Nearctic areas into East Nearctic (EN) and West Nearctic (WN), a pattern commonly observed, where EN and WN are often more associated with Palaearctic areas than between each other. However, considering that the barbata-group is a strictly Nearctic clade, as far as we know, it is valid to consider Halffter’s (1987) proposition, in which this distribution pattern (EN and WN) is recurrent in recently speciated insect groups, suggesting time h2, where the achradoceran species have undergone a fairly recent radiation.

In South America, the divisions observed between transandine (= west of Andes) species, such as A. meridionalis and A. angustifacies, and cisandine (= east of Andes) species, such as A. longiseta, A. insignis, A. tuberculata, A. gimli, and A. balin, indicate a strong Andean influence on the genus distribution. Andean radiations are complex to analyze; geological reconstructions indicate that uplift events occurred at various periods throughout time, advancing from south to north and from west to east (Hoorn et al. 1995; Gregory-Wodzicki 2000; Taylor 1991; Garzione 2008). Following the interpretation of time h2, the vicariant event recovered in clade D must be related to some more recent uplifting, followed by subsequent isolation in Chile of the species A. angustifacies, possibly through an event related to the South American Transition Zone (SATZ) (Morrone 2004). The transandine recolonization by A. tuberculata may have also been caused by some vicariant event related to SATZ, for clade J’ in T2 (Fig. 4)–we consider that founder events are not viable for a more detailed explanation, due to multiple factors and stochastic events that may be involved, so most of them will not be addressed. This is also the case for the founder event hypothesis of A. tuberculata in clade J in T1, but also in H or L for A. femoralis and A for A. longiseta. Although it is not possible to directly detect their causes, founder events–which promotes speciation through a rapid bottleneck effect resulting from a small, and isolated population (Barton, 1984)–have been widely supported in biogeographical studies (Matzke, 2014). The founder event was incorporated into the method of Arias (2017) based on the following principle: one descendant inherits the whole ancestor’s range, and the other descendant starts as a founder population outside the ancestral range.

The vicariant event recovered for A. excavata is controversial. The current position of the Greater Antilles was reached in the Miocene (about 25 Myr), when the Caribbean Plate collided with the Bahamas Block (Buck 1990). These islands have no recent connection with continental landmasses, having been separated from South America in the Cretaceous (Anderson and Schmidt 1983). Nevertheless, due to fluctuations in sea level, it is difficult to determine exactly how much land remained above water during prehistoric times, if any (Buck 1990). Therefore, we can speculate about some land connection between Cuba and Mexico that might justify this result. However, it is also plausible that it could be some analytical noise in the GEM, and a founder event, like the one that originated A. apicalis in the Lesser Antilles, may have also promoted the speciation of A. excavata. One plausible explanation for these cases could be dispersal through air masses or surface debris (Bickel 1996). An alternative hypothesis for the current distribution pattern of A. excavata and A. apicalis is the GAARlandia hypothesis (GAAR = Greater Antilles + Aves Ridge), introduced by Iturralde-Vinent and MacPhee (1999), but which has gained strength recently (Alonso et al. 2011). In this hypothesis, land masses connected South America with the Caribbean Islands about 34 Myr, resulting from a combination of tectonic compression factors and a rapid sea-level fall, due to a fast ice-sheet growth in Antarctica (Ali 2012). However, taking this hypothesis into account pushes the time of cladogenesis of Achradocera even further in the past. In time, stating anything would be premature, especially since we lack “branching clock” data, let alone a minimum fossil record. Nevertheless, considering that this is a seminal work for the biogeographical patterns of Dolichopodidae in the New World, we believe that raising some historical possibilities may stimulate future discussions in the field.

In general, the lowland Pan-American Dolichopodidae fauna commonly presents species with wide distribution areas, disregarding “areas” traditionally used for biogeographical studies (Morrone 2017). Thus, biogeography methods that do not rely on areas (Hovenkamp 1997, 2001, 2002; Arias et al. 2011) and, alternatively, emphasize the processes generating cladogenesis, may be potentially favorable for future research with this taxon. Despite being recent and still presenting evident limitations, GEM (Arias 2017) has proven to be a very promising tool. Finally, this study was only possible thanks to a previous and extensive review of the genus (Quevedo et al. 2024) and the morphological study that provided the phylogeny, thus reducing Linnaean, Darwinian, and Wallacean shortfalls. Still, some data noises were unfeasible to resolve for now, we highlight two: (1) the lack of systematic knowledge about Chrysotus made it not possible to confidently choose a sister group for Achradocera for the biogeographical analysis, creating noise in the reconstruction of the genus’s ancestral distribution; (2) collection bias, where many areas in the interior of Brazil and entire countries in South and Central America, lacked occurrence records.