A total of 223 specimens were examined belonging to the following institutions: ARC – Arthropod Research Collection, Michigan State University, Michigan, USA; DZUP – Coleção de Entomologia Pe. Jesus Santiago Moure, Curitiba, Brazil; MNCR – Coleção de Entomologia Pe. Jesus Santiago Moure, Curitiba, Brazil [formerly Instituto Nacional de Biodiversidad – INBio], Santo Domingo de Heredia, Costa Rica; MZSP – Museu de Zoologia da Universidade de São Paulo, São Paulo, Brazil; NHMUK – Natural History Museum, London, England; ZMHB – Berlin Museum für Naturkunde der Humboldt-Universität, Berlin, Germany. Other repositories cited in the text are: NMSA – Natal Museum, Department of Arthropoda, Pietermaritzburg, South Africa; and USNM – National Museum of Natural History, Washington DC, USA. Specimens collected in the Brazilian states of Mato Grosso, Mato Grosso do Sul and Rondônia from the SISBIOTA-Diptera Project (CNPq-FAPESP), coordinator Carlos Lamas, vice-coordinator Silvio Nihei, were also examined.

To study adult morphology, dried and pinned specimens were examined under a Leica EZ4 stereomicroscope. A Leica DM2500 optical microscope was used to analyze the postabdomen, first instar larvae and spermathecae.

To study the male postabdomen, the specimens were carefully dissected from the fifth segment to avoid damaging the sixth tergite and to maintain the integrity of the abdomen as much as possible. To study the female abdomen and obtain the spermathecae, first instar larvae and/or eggs, the abdomen was dissected from the fourth segment and rarely in the third. The male terminalia were bleached in 10% potassium hydroxide solution (KОН) for four minutes in boiling water, neutralized with 5% acetic acid solution and washed with water. The female terminalia, larvae, spermathecae and eggs were subjected to a similar procedure, except they were heated for 10 minutes in 10% KOH solution. At the end of the procedure, the material was preserved in glycerin, packed in microplastic vials and pinned with the respective specimen.

The terminology of adult and spermathecae morphology follows Cumming and Wood (2017). The terminology used for the wing structure and trace of M₂ vein is taken from Crosskey (1976). For male terminalia, we follow Tschorsnig (1985). The terminology of the first instar larva follows Thompson (1963), with some modifications discussed by Cantrell (1988). The term “cephaloskeleton” from Courtney et al. (2000) was used. The terminology for the puparium follows Ziegler (1998) and that for the eggs follows Gaponov (2003).

To select the terminals of the ingroup, three premises were considered: (1) the availability of adult specimens for morphological study; (2) the availability of immature stage material (e.g., first instar larvae); and (3) differences in geographic distribution and morphology. All 13 genera included in Dufouriini (sensu Herting 1984) (Table 1) were studied, and 11 were sampled, including those genera from the formerly valid Oestrophasiini. Additionally, the sole genus currently assigned to Freraeini (Freraea) (O’Hara et al. 2020) was included, as it has historically been related to Dufouriini. The ingroup included 26 species from 13 genera. In light of the results of Cerretti et al. (2014) (where Dexiinae is paraphyletic in relation to Phasiinae) and Stireman et al. (2019) (Dufouriini polyphyletic, split into two lineages), some additional representative tribes were chosen as outgroup taxa. The basis of paraphyly of Dexiinae derived from historically problematic taxa: Strongygastrini, Imitomyiini and Catharosiini, besides other Phasiinae (Cylindromyia) were included. We also included species of Dexiini and Voriini concerning Dufouriini monophyly and relationships. Xanthozona (Tachinini) was selected as the root for the analyses. The outgroup included a total of nine species from Phasiinae, Dexiinae and Tachininae. Supplementary file 1 shows the terminals included in the cladistic analysis with geographical distribution, data source and observed structures (whether personal observation or literature data).

The study of phylogenetic relationships was based on morphological characters of adults (including female and male genitalia and spermathecae), first instar larva, egg and puparium, which was based on parsimony as the optimality criterion. Whenever possible, the characters were constructed according to the proposal of Sereno (2007), with preference for the contingent coding (Forey and Kitching 2000). Characters are scored with “–” in case of inapplicability (usually taxon lacking the character-bearing structure), and with “?” in case of lacking observation. The data and putative synapomorphies of the male terminalia presented by Tschorsnig (1985) were reanalyzed and included within a cladistic framework. Characters from the literature, e.g., Cerretti et al. (2014), have been properly indicated in the character list.

The polarization was conducted using the method of outgroup comparison (Nixon and Carpenter 1993). The matrix of characters was built with Mesquite 3.04 software (Madison and Madison 2015). For the parsimony analysis using equal and implied weighing, the TNT 1.1 software (Goloboff et al. 2008) and the strategies of the New Search Technology (Ratchet, Drift, Tree Fusion and Sectorial Searches) were used. The analysis was performed according to the following parameters: random seed = 1; number of replicates = 10,000; number of trees saved per replication = 10. The software Winclada 1.00.08 (Nixon 2002) was used to display the trees with the transformation series of each character, in addition to its optimization. For the MP tree under equal weights, we provide the total length (L), the consistency index (CI) (Kluge and Farris 1969) and the retention index (RI) (Farris 1989), calculated from all characters.

The parsimony criterion of Fitch (1971), which treats the characters as unordered (or non-additive), was used in this study. Autapomorphic characters of single terminals were maintained in the analysis because they are part of cladistic results (Yeates 1992). Implied weighting (Goloboff 1993) was used to observe how the characters behave in different weighing schemes, based on the fit measure of each character and its overall fit of the topology. The k-values of 1, 2, 3, 5 and 10 were tested. Branch support was checked using Bremer support (1994), with the “Bremer.run” script provided in the TNT Software Wiki (http://phylo.wdfiles.com).

Character optimization is often performed following the proposal of De Pinna (1991), which argues that ACCTRAN is preferable to DELTRAN because it preserves more primary homology hypotheses of. However, Agnarsson and Miller (2008) argue that they do not see theoretical components that make ACCTRAN more preferable than DELTRAN. Amorim (2002) argues that it is more reasonable to analyze the evolution of the characters case by case and to explicitly explain the reason for using ACCTRAN or DELTRAN rather than using only one optimization for all characters. Thus, in some cases (e.g., when there are terminals with non-observable or inapplicable state) ACCTRAN would consider it a spurious synapomorphy, whereas DELTRAN does not perform this transformation, considering an apomorphy for the taxa that have the given state only. Thus, it is safer to adopt the latter. The preference of each optimization was explicitly indicated in the character list.

Most characters were illustrated using photographs and line drawings to facilitate identification of different character states. The photographs were taken with a Leica DFC420 digital camera coupled to a Leica MZ16 stereomicroscope. The images were obtained through the software LAS V4.1, then stacked in the software Helicon Focus 5.3.14 and edited in the software Adobe Photoshop CS6 and Adobe Illustrator CS6. The microphotographs of eggs and puparia were processed in Balzers CPD 030, and later were metallized in the Balzers SCD 050 for analysis using the scanning electron microscope, Zeiss DSM 940. In addition, drawings were made using the Leica DM2500 optical microscope with its coupled camera. Subsequently, these drawings were vectored and edited in Adobe Illustrator CS6 software.

A total of 185 characters were constructed, 5 of the egg, 22 of the first instar larva, 7 of the puparium (posterior spiracle), 67 of the external morphology (except terminalia) 53 of the male terminalia, 23 of the female terminalia, and 8 of the spermatheca. The optimizations of the ambiguous characters will be discussed. When relevant, comments will be made for some characters. The characters from literature will be properly referenced with the statement of the author and/or first observer.

Our study included 35 species and 22 genera, with 26 species and 13 genera in the ingroup. All genera of Dufouriini (including Oestrophasiini and excluding Mesnilana and Rinophoroides, see more in Discussion) and Freraeini were sampled. Our holomorphological analysis included a total of 185 characters from the egg (5 characters), first instar larva (22), puparium (7), adult external morphology (67, excl. terminalia), female terminalia (23), male terminalia (53) and spermatheca (8). The data matrix is provided in Supplementary file 2.

Cladistic analysis with equal weights resulted in a single, most parsimonious tree (L = 400; CI = 61; RI = 83) (Fig. 23). The implied weighting analysis resulted in a single tree with the same length and topology as the equal weighted analysis, but with differences in the optimization of some characters. The single most parsimonious tree with equal weighting will be used in the discussion with unambiguous characters optimized and clades numbered (Fig. 24). Cladograms with ACCTRAN and DELTRAN character optimization, in addition to the Bremer support of each clade, are provided in Supplementary file 3.

Dufouriini, as defined prior to this study (Table 1), was considered paraphyletic. In the present analysis, the genera Microsoma and Pandelleia form a clade with (Freraea + Eugymnopeza), supported by eight synapomorphies (clade 6), constituting the new definition of the tribe Freraeini. Thus, Eugymnopeza, Microsoma and Pandelleia, prior to this analysis as Dufouriini, were recovered in a clade within Freraeini; while the remaining genera of Dufouriini, i.e., composed only of Rondania, Chetoptilia, Dufouria, Comyops and Ebenia clustered in their own clade (clade 13), i.e., Dufouriini. The former Oestrophasiini genera Cenosoma, Euoestrophasia, Jamacaria and Oestrophasia are grouped in a strongly supported clade defined by 19 unambiguous synapomorphies (clade 10). This clade is sister group to Dufouriini s.s. (clade 13), and defined by three synapomorphies. Based on these results, we are here considering Oestrophasiini as a valid tribe separate from Dufouriini.

Dufouriini s.s. (clade 13) as here defined is composed by five genera: Rondania, Chetoptilia, Dufouria, Comyops and Ebenia. It is supported by three synapomorphies: antennae with micropubescent arista (55:1); spermathecae with pores (179:1); and male terminalia with distiphallus with anterior margin partially sclerotized (142:2 under DELTRAN); and one homoplasy: female terminalia with elongate sternite 8 (172:2).

In the internal resolution of Dufouriini s.s., Rondania is sister group to the clade grouping all other genera. This clade (14) is supported by seven synapomorphies: first instar larva with segment I with flattened antenna (9:0 in DELTRAN); conical segment XII (16:1 in DELTRAN); sclerite of the salivary gland narrow anteriorly and wide posteriorly (20:1 in DELTRAN); accessory sclerite falciform (21:4); intermediate region with median enlargement (26:1 in DELTRAN); female terminalia with tergite 8 fused with sternite 8 (168:1); sternite 10 sharp and curved (174:1). Chetoptilia is sister group to (Dufouria (Comyops + Ebenia) (clade 15), supported by four unambiguous synapomorphies: fronto-orbital plate with several setae on the antennal socket (43:1); male terminalia with phallapodeme with fan-shaped apex (126:1); distiphallus with ventral sclerite dorsal projection (139:1) and distiphallus with distal portion (144:1) and two unambiguous homoplasies. Dufouria is sister group to clade 16 (Comyops + Ebenia) supported by one unambiguous synapomorphy: male terminalia with surstylus with lateral setae (119:1), and four unambiguous homoplasies. Along with Comyops, we also sampled Comyopsis, represented by its type species, C. fumata, which is sister group to Ebenia species supported by one unambiguous synapomorphy, and therefore Comyopsis is here synonymized with Ebenia (see discussion below).

Dufouriini s.s. is sister group to Oestrophasiini (clade 9) based on seven unambiguous synapomorphies: epandrium with lateral lobes (111:1); epandrium with posterior projection zone (112:1); phallapodeme larger than hypandrium (127:1); basiphallus dorsally segmented (132:1); ejaculatory apodeme fan-shaped (146:1); anterior margin of postgonite with weak sclerotization (153:0); asymmetric spermathecae (183:1). Oestrophasiini (clade 10) as here defined and revalidated is formed by Cenosoma, Euoestrophasia, Jamacaria and Oestrophasia, as recognized by Guimarães (1977), and supported by 19 unambiguous synapomorphies: microtype egg (1:2); first instar larva with antena present, but reduced (8:2); segment II with spines with triple development in relation to length of adjacent microtrichia (13:1); posterior spiracle with conical felt chambers (19:1); sclerite of salivary gland rounded (20:2); puparia with completely fused peritreme (30:1); spiracular openings arborescent (33:1); fronto-orbital plate yellow in males (41:2); face with setulae on lunule (51:1); scutum yellow with black spots (66:1); femur II with 3 submedian anterodorsal setae in females (79:2); wing membrane with macules (80:1); tergite 5 with pair of dark brown rounded spots on ventral posterolateral region (100:1); yellow tegument (101:1); male terminalia with tergite 6 fused, but with distinguishable limits (from lateral prominences) segment 7 + 8 (103:3); pregonite fused together with downwards directed apex (151:1); female terminalia with bare tergite 6 (158:0); syntergosternite 7 ring-shaped (165:1); tergite 8 with narrow plate shape (167:1); besides seven unambiguous homoplasies.

Freraeini, by including Pandelleia, Eugymnopeza and Microsoma, formely in Dufouriini, provide evidence of this newly delimited clade as monophyletic (clade 6). This redesigned tribe is composed and related as follows: (Pandelleia (Microsoma (Freraea + Eugymnopeza). Here, this newly delimitation of Freraeini is supported by eight unambiguous synapomorphies: first instar larva with spines on segments I–XII (12:1); rectangular sclerite of salivary gland (20:3); triangular accessory sclerite (21:2); mouthhook with same basal thickness as dorsal cornu (23:1); accessory sclerite with ventral position with regard to sclerite of salivary gland (25:1); dorsal cornu shorter in length compared to mouthhook (27:1); vertex with protuberant ocellar triangle not protuberant (38:1); and female terminalia with tergite 8 fused with sternite 8 and 9 (170:1); in addition to five unambiguous homoplasies.

All members of Dufouriini s.s., Oestrophasiini and Freraeini form a monophyletic clade (clade 5) outside the Dexiinae and sister group to the Phasiinae exemplars included herein (clade 3). Clade 5 is supported by four unambiguous synapomorphies: anepimeron with fine setae (77:2); male terminalia with hypandrial apodeme with boundary with central plate indistinct (123:2); distiphallus with extension of dorsal sclerite more than half length of median bar (137:1); and female tergite 8 elongated, when free (167:3), in addition to three homoplasies: holoptic male with dichoptic female (35:1); female terminalia with syntergosternite 7 (tergite 7 fused with sternite 7) present (164:1) and sternite 10 reduced (174:3).

The Phasiinae was recovered as sister group (clade 4) of the tribes Dufouriini s.s., Freraeini and Oestrophasiini (clade 5), being supported by six synapomorphies and two homoplasies.

Dufouriini was recovered as paraphyletic, confirming earlier results by Ziegler (1998), Barraclough (‎2005), Cantrell and Burwell (2010) and Cerretti et al. (2014); except by Stireman et al. (2019) that recovered Dufouriini as polyphyletic. A broad definition of Dufouriini including Microsoma and Pandelleia, as well as the four genera of Oestrophasiini, was not supported here. Our analysis supports splitting the previous delimitation of Dufouriini, i.e. sensu lato (Table 1), into three strongly supported and closely related tribes (clade 5): Dufouriini s.s. (hereafter, just Dufouriini), Oestrophasiini and Freraeini. Dufouriini is composed of five genera and defined by the synapomorphies mentioned above in the Results section.

Cerretti et al. (2014) considered Dufouriini in the broadest sense, comprising all the genera from Dufouriini, Oestrophasiini and Freraeini. Their six Palaearctic genera sampled (Chetoptilia, Dufouria, Rondania, Pandelleia, Freraea and Eugymnopeza) were paraphyletic and graded with a monophyletic Phasiinae. Stireman et al. (2019), on the other hand, considered Dufouriini (incl. Oestrophasiini) and Freraeini separately. The former was sampled with five genera (Oestrophasia, Rondania, Microsoma, Dufouria and Ebenia), while the latter with one (Freraea). Their recovered Dufouriini (with four genera of Dufouriini and Oestrophasiini) and Freraeini (with two genera) as not closely related, but instead intergraded by Telothyriini and by small clades of Voriini and Palpostomatini.

Although our analysis was based on a complete generic sampling of Dufouriini, Oestrophasiini and Freraeini and considered a comprehensive and detailed morphological study of adult and immatures stages (totaling 185 characters), our results might be limited, especially concerning supratribal relationships. On one hand, those three tribes were strongly supported by comprehensive morphological evidence and based on thorough sampling of each tribe. On the other hand, to obtain a reliable intertribal relationship, a more comprehensive sampling of other tribes of Dexiinae (and perhaps Phasiinae) is recommended and desired. Our outgroup sampling was composed of taxa of Phasiinae, that were found to be closely related to Dufouriini (Cerretti et al. 2014), and Dexiinae, wherein Dufouriini are considered to belong, so it is expected that these closely related taxa to Dufouriini present the greatest potential to access the robustness of its monophyly (Grant 2019). Thus, we believe this sampling was sufficient for establishing the monophyly of Dufouriini, Oestrophasiini and Freraeini, as we are not inferring its placement within Tachinidae. Finally, we are confident that our choice of outgroups provides a crucial test of the ingroup topology – by evaluating the ingroup character-state transformations (Grant 2019) – as it can reliably answer our question within this paper, i.e., what are the relationships among the genera and the tribes Dufouriini, Oestrophasiini and Freraeini.

Given the size, diversity and distribution of Tachinidae, taxonomic sampling in Stireman et al. (2019) was far from complete, but was enough to shed light on several questions. In this sense, their findings were a step forward since they indicated that the delimitation and relationships of Dexiinae groupings remains unclear and puzzling. Now, as it will be discussed, our study adds some more evidence to the classification of Tachinidae by supporting that Dufouriini is not a single tribe, but, in fact, three separate tribes.

Herting (1957, 1960) grouped the taxa with modified ovipositor (syntergite 9 + 10) in Dufouriini s.l., composed of the following Palaearctic genera: Chetoptilia, Dufouria, Eugymnopeza, Freraea, Microsoma, Pandelleia and Rondania. However, the three synapomorphies and one homoplasy for the tribe found herein (clade 13) were not from the female terminalia. Besides, the homology among their ovipositors was not conclusively demonstrated (O’Hara and Wood 2004), and in the present analysis some structures were considered non-homologous. For example, Rondania has a posteriorly directed tube-shaped ovipositor, completely fused syntergosternite 6 + 7 and lacks sternite 9, while in Freraea and Eugymnopeza the tube-shaped ovipositor is directed anteriorly, has a partially fused syntergosternite 6 + 7 and well-developed sternite 9.

The configuration of genera recovered here highly agrees with Verbeke (1962), with the Dufouria group within his Dufouriines, containing the Palaearctic genera Chetoptilia, Dufouria and Rondania. Accordingly, herein, Rondania is sister group to the genera Chetoptilia, Dufouria, Comyops and Ebenia. The last two genera mentioned, Comyops (including Comyopsis) and Ebenia, pertained to the Neotropical tribe “Ebeniini”. This tribe, currently invalid and formerly composed of 11 genera, was an assemblage of many unrelated taxa that was put together by Townsend (1936). The remaining genera of the former tribe Ebeniini of Townsend (1936) are currently placed in different tribes, like Voriini, and even subfamilies, like Palpostomatini in Tachininae (O’Hara et al. 2020). However, when better studied, some of the “Ebeniini”, i.e., Ebenia, Comyops and Comyopsis, showed affinities with the Dufouriini as discussed by Thompson (1963) and placed formally in Dufouriini by O’Hara et al. (2020). Thompson (1963) argued for a probable relationship between Comyops, Comyopsis and Ebenia with Dufouria based on larval anatomy and cephaloskeleton (similar to Dufouria chalybeata Meigen), as well as male terminalia (Comyopsis resembling Dufouria occlusa (Robineau-Desvoidy, 1863)). This relationship was confirmed here with Comyops and Ebenia as sister group to Dufouria (clade 15). Additionally, Comyopsis is considered a junior synonym of Ebenia herein (see below).

Mesnil (1975) delimited Dufouriini into three subtribes: Dufouriina with Dufouria; Campogastrina with Chetoptilia, Pandelleia, Rondania and Microsoma; and Freraeina with Eugymnopeza and Freraea. His classification was not recovered herein, with some genera of Campogastrina placed in Dufouriini (Chetoptilia, Rondania) and others in Freraeini (Pandelleia and Microsoma). Ziegler (1998) assigned Rondania and Dufouria to Voriini s.l. based on the 3rd instar larva cephaloskeleton, along with Stireman et al. (2019), where Dufouriini taxa were graded within Voriini (in addition to Palpostomatini and Telothyriini). O’Hara and Wood (2004) characterized members of Dufouriini (Dufouria, Rondania and Oestrophasia) as having a fused pregonite, thus setting apart Microsoma and Freraea. This character (150:1) was confirmed as an ambiguous homoplasy grouping Dufouriini and Oestrophasiini (clade 9). Before comparing our results with the phylogenetic hyphotheses of Cerretti et al. (2014) and Stireman et al. (2019), it is worth noting that recently O’Hara et al. (2020) placed Kambaitimyia Mesnil, 1953 in Dufouriini. This genus, known from two species from Myanmar, was originally assigned to Dufouriinae (Dufouriini, in part) by Mesnil (1953). However, later he changed his mind (Mesnil 1966) and placed this genus within his subtribe Ptilopsinina near Macquartini and Leskiini in Tachininae, only to be placed again in Dufouriini by Crosskey (1976) - by relying only on the similar external adult facies with other taxa placed in this tribe by him, including the Macquartini (Tachininae) genus, Anthomyiopsis Townsend, 1916. Verbeke (1962) was the first author who examined the male terminalia of Kambaitimyia (K. carbonata Mesnil, 1953), and concluded that the presence of a reduced distiphallus inserted on an U-shaped basiphallus is very close to the Strongygaster group and it would be best placed in a group including genera like Imitomyia, Strongygaster and Rondaniooestrus (all currently placed in Phasiinae). Later, Tschorsnig (1985) confirmed Verbeke’s (1962) conclusion and placed Kambaitimyia in Strongygastrini (Phasiinae). Herein, by examining and dissecting a male of K. carbonata from NHMUK, we further confirm the peculiar and strong resemblance of the male terminalia of members of the genus Strongygaster and confirm the conclusion of Verbeke (1962) and Tschorsnig (1985) that Kambaitimyia is conclusively not a Dufouriini and is probably best placed in Strongygastrini.

The results of Cerretti et al. (2014) with Dufouriini s.l. as paraphyletic and closely related to Phasiinae was partially confirmed herein. Our analysis confirms the close relationship between Dufouriini s.l. and Phasiinae, but both were monophyletic and sister groups. The clade with Dufouriini s.l. + Phasiinae was supported by a single homoplasy (character 45:0 of Cerretti et al. 2014): presutural acrostichal seta absent; which is not a reliable character, since it appears independently in several other taxa within Tachinidae and other muscoestroid families. Besides, the genera that were restricted to Dufouriini did not group together. Stireman et al. (2019) recovered part of Dufouriini s.l. forming a clade with (Ebenia (Dufouria (Rondania + Oestrophasia), which was confirmed here, but included a less comprehensive sampling. We support both Dufouriini and Oestrophasiini as monophyletic and sister group to each other (clade 9).

Our results diverge from Herting (1957, 1960, 1984), who brought together Eugymnopeza and Freraea in Dufouriini based on the structure of the ovipositor, thereby invalidating Freraeini. Herein, Freraeini (clade 6) was supported by two ovipositor characters: one synapomorphy (tergite 8 fused with sternites 8 and 9 (170:1)) and one homoplasy (sternite 8 elongated (172:2)). On the other hand, Freraeini defined herein agrees partially with Verbeke (1962) and his Freraea-group (containing Freraea, Litophasia and Microsoma) and Pandelleia-group (with Pandelleia). Verbeke based these groups on male terminalia: the former group has thin and elongated distiphallus and the latter a reduced and subrectangular basiphallus. We obtained two homoplasies from the male terminalia supporting Freraeini: tergite 6 fused but with visible suture (median dividing line present) in segment 7 + 8 (103:2) and long basiphallus (133:0). Excluding Litophasia (see below), Verbeke’s (1962) proposal to leave these genera outside the Dufouria-group was accurate according to the present study, because Dufouriini (clade 13) does not include Freraea, Eugymnopeza, Pandelleia and Microsoma. These three genera in turn belong to Freraeini (clade 6), similar to his Pandelleia-group plus Freraea-group (except for Eugymnopeza, which Verbeke did not study).

Mesnil (1975) considered his subtribe Freraeina with the same genera as Townsend (1936), with Freraea and Eugymnopeza only. Herein, this subtribe was monophyletic (clade 8). He commented that Microsoma is very closely related to Freraeina, and we confirm here Microsoma as sister group of Freraea + Eugymnopeza (clade 7). O’Hara and Wood (2004) transferred Freraea from Dufouriini to Freraeini. Later Eugymnopeza was too placed in Freraeini by O’Hara et al. (2009), agreeing with Townsend (1936) and Mesnil (1975), a relationship that was confirmed in the present study; however, in clear contrast, O’Hara et al. (2020) changed the placement of this genus one more time, and returned it to Dufouriini. Furthermore, the character used for this transfer, presence of fused pregonite (our character 150:1), was confirmed as a synapomorphy for clade 9 (Oestrophasiini + Dufouriini). Cerretti et al. (2014) recovered a clade with most Freraeini genera ((Pandelleia + Rondania) (Microsoma (Eugymnopeza + Freraea)))) supported by one character from the female terminalia, tergite 6 long and tubular (Cerretti et al. 2014, character 128:1). However, when scrutinized, this character shows differences among these genera. Although both Pandelleia and Rondania have a long and tubular tergite 6, it is anteriorly directed (160:0) in Pandelleia, while it is posteriorly directed (160:1) in Rondania. Additionally, only Rondania possesses fully telescoped terminalia. The relationships found by these authors is nearly identical to those found herein, differing only by the presence of Rondania, which was placed in Dufouriini herein (clade 13). Finally, the presence of the six unique synapomorphies of the first instar larva (as listed in Results), in addition to the unique synapomorphy found on the female terminalia – tergite 8 fused with sternites 8 and 9 (170:1) – are compeling evidence for the unique habit of host infection that evolved in Freraeini. As this tribe, in the same way of Oestrophasiini and Dufouriini, attacks adult Coleoptera, the functional solution to overcome this challenge was developed by some of its members (Pandelleia, Eugymnopeza and Freraea). Thus, they place the eggs inside the beetles with their terminalia in order to infect them; Microsoma, distinctively, avoided this problem by piercing the sclerite of the beetle with its sharp terminalia. This strategy, even if functionally equivalent to some Dufouriini (Chaetoptilia, Dufouria, Ebenia and Comyops), is morphological different in Microsoma, particularly the larva and the female terminalia, as it happens to the other members of Freraeini. It differs considerably from those genera of Dufouriini as pointed by the synapomorphies above, and clearly indicate a unique solution to infect their hosts. Thus, our preference to maintain this tribe as unique and separate from Dufouriini (clade 13).

The clade (Microsoma (Eugymnopeza + Freraea) of Cerretti et al. (2014) supported by one synapomorphy (anteriorly curved tergite 5, character 126:1) was also recovered here (clade 7) and supported by five synapomorphies. Eugymnopeza + Freraea was supported by two homoplasies in Cerretti et al. (2014): ocellar seta present or absent (polymorphic) (character 17:1/2) and short chaetotaxy, thin and reclined bristles that covers most of its surface or reaching at least the lower half (character 24:4). It was recovered here (clade 8) with one unambiguous autapomorphy and three homoplasies. Accordingly, Stireman et al. (2019) recovered Freraea as sister group of Microsoma, but Eugymnopeza was not sampled.

Litophasia is a very special case, as it has been considered in Catharosiini (Phasiinae) (Cerretti et al. 2014) and recently as an unplaced Dexiinae (Blaschke et al. 2018; Stireman et al. 2019). While a definitive placement for Litophasia is unknown, some clues might signal a possible relationship with Freraeini. Moreover Stireman et al. (2019) have indicated Litophasia as close to Freraea.

Guimarães (1971) considered the Neotropical genera of Glaurocarini sensu Townsend (1936) as the new tribe Oestrophasiini. Moreover, Mesnil (1973) mentioned that Townsend (1936) erroneously classified Oestrophasia and Cenosoma in Glaurocarini and these genera are related to Dufouria, with a connection with his subtribe Campogastrina near Chetoptilia. Our results partially agree with Mesnil’s (1973) since members of his Campogastrina, namely, Chetoptilia, Pandelleia and Rondania, but not Microsoma, are placed in Dufouriini and sister group to Oestrophasiini (clade 9). Guimarães (1977), in his revision of Oestrophasiini, discussed a likely relationship of this tribe with the Old World Dufouriini based on Verbeke’s (1962: pl. X) genitalia drawings of Chetoptilia, Dufouria and Rondania.

Tschorsnig (1985) formally considered Oestrophasiini as belonging to Dufouriini. O’Hara and Wood (2004) agreed with Tschorsnig (1985) based on the presence of a fused pregonite. This character was used here (character 150:1) and appeared to be an ambiguous homoplasy grouping Oestrophasiini and Dufouriini. Despite the importance of the pregonite, a relevant synapomorphy for Oestrophasiini is the presence of microtype eggs (character 1:2). Thus far, this feature had only been considered to be present in Goniini and some Blondeliini (Gaponov 2003), however we found and characterized it as present in Oestrophasiini based on the evidence provided by Gaponov (2003) and Salked (1980) for the eggs, the internal morphology of the female and the larva by Thompson (1924, 1963). This is so because these eggs are very small in size (less than 0.4 mm in length); are placed on leaves and are accidentally ingested by the host, which are thus infected (Grillo and Alvarez 1984); are present in high quantity (between 2,000 and 3,000); the female ovary have more than 100 ovarioles (Grillo and Alvarez 1984); while the larvae have extremely reduced antennae and posterior spiracles; transparent and colourless cuticle, with rows of spines at the posterior end of the first two thoracic segments; segment I extremely well-developed and pigmented, with the rest of the body without spines. Accordingly, the important biological significance of the presence of microtype eggs in Oestrophasiini, which indicates a very specific and complex adaptation to host infection (Gaponov 2003; Thompson 1963), in addition to the posterior spiracles of the puparia, with the peritreme completely fused (character 30:1) – constituting a unique characteristic within Tachinidae, unknown elsewhere in the family (Ferrar 1987; Greene 1921; Ziegler 1998) – confirm that this tribe is best ranked as a separate tribe from Dufouriini. Moreover, an additional 17 unambiguous synapomorphies are shared by Oestrophasiini and separate them from Dufouriini.

Still within Oestrophasiini, Wood (1987) synonymized Cenosoma with Oestrophasia, an act that was maintained by O’Hara and Wood (1998, 2004). Here the synonymy was not supported, with Oestrophasia monophyletic and supported by four autapomorphies and two homoplasies, and sister group of Cenosoma, Euoestrophasia and Jamacaria (clade 11). Based on this evidence, Cenosoma and Oestrophasia are considered as distinct genera herein. Jamacaria is a monotypic genus that is sister group to Cenosoma. Finally, our analysis did not support the placement of Cenosoma thompsoni as unplaced species of Oestrophasia (sensu O’Hara et al. 2020) as done by O’Hara et al. (2020). Contrarily, our phylogenetic analysis places Cenosoma thompsoni conclusively within Cenosoma as proposed by Guimarães (1977).

Herein, Comyopsis Townsend, 1919 is conclusively transferred from the former tribe Ebeniini to Dufouriini, confirming the proposal of Thompson (1963), and most recently by Stireman et al. (2019) and O’Hara et al. (2020). Additionally, following our phylogeny, we also propose Comyopsis as a junior synonym of Ebenia Macquart, 1846. Furthermore, our work does not confirm the proposition of O’Hara et al. (2020) that, oddly, placed Comyopsis in Voriini. In our analysis however, Comyopsis fumata is sister group of E. claripennis + Ebenia sp. 1 (within clade 16). Unlike Townsend’s (1927: 234) key, C. fumata does have a costal spine and vein R_{4 + 5} with setulae reaching crossvein r-m, as well as Ebenia species. In Thompson’s key (1963: 342), the couplet separating Ebenia and Comyopsis uses the length of the costal spine (long in Comyopsis, short in Ebenia) and wing membrane pigmentation (smoky in Comyopsis, and totally hyaline in Ebenia). After examining some species of Ebenia, we found that the only characteristic distinguishing these genera is the setulose prosternum in Ebenia. We considered this character as very unsubstantial to justify generic separation. Besides, there is no significant difference between their male terminalia, therefore, we propose a synonymy between Comyopsis and Ebenia. The only species of Comyopsis, C. fumata Townsend, 1919 (type-locality: Nicaragua, Chinandega) is consequently transferred to Ebenia. However, when O’Hara et al. (2020) placed the previously unplaced species of “Ebeniini” (Guimarães 1971), Ebenia fumata (van der Wulp, 1891) in Ebenia, our new synonymy, E. fumata (Townsend, 1919), constitutes a junior secondary homonomy. In order to resolve this issue, we herein propose a new name for this new combination: Ebenia neofumata Santis and Nihei nomen novum for Ebenia fumata (Townsend, 1919) [nomen preoccupatum].

The Afrotropical genera Mesnilana, with one single species M. bevisi Emden, 1945, and Rhinophoroides, also with one single species R. minutus Barraclough, 2005, were originally included in Dufouriini. Emden (1945) erected Mesnilana for a female from South Africa and included it by considering the classification of Mesnil (1939), which was then in Phasiinae. In the generic description, he commented (1945: 414): “The longer antenna would seem to approach this genus to the Ocypterini [Cylindromyiini, in part], but the general appearance, genitalia, dark occipital hairs, etc. make it more closely related to Diplopota [= Imitomyia Townsend]”. Thus, Imitomyia, currently placed in its own tribe (Imitomyiini) in Phasiinae (Tschorsnig 1985) or uncertain position (Stireman et al. 2019), would be the closest genus of Mesnilana. Later, Crosskey (1980, 1984) maintained Mesnilana in Dufouriini, but in the subfamily Dufouriinae. Barraclough (2005) described the new genus Rhinophoroides and placed it in Dufouriini and subfamily Dexiinae, because of its great resemblance with Mesnilana. Actually, Barraclough reported that he did not observe any close relationships between any other Afrotropical genera of Dufouriini and included Rhinophoroides in this tribe by relying only on the general similarity with Mesnilana. In the Afrotropical Catalogue (O’Hara and Cerretti 2016), Mesnilana was tentatively placed in Dufouriini and it was pointed that Rhinophoroides could be a junior synonym of Mesnilana.

The female holotype of Mesnilana bevisi deposited at NHMUK was recently examined by MDS, and by carefully observing the descriptions provided by Barraclough (2005), we found that the observed features do not correspond to the Dufouriini as redefined herein, nor with any of the related tribes, Freraeini and Oestrophasiini. Some of these characters include the bare facial ridge, three katepisternal setae and anepimeron with a well-developed seta. In addition to external morphology, more evidence seems to provide an important biological insight: both genera were collected in light traps, suggesting nocturnal hosts (Barraclough 2005). This is not known from other members of Dufouriini and is uncommon in Tachinidae (occurring, for instance, in the cricket parasitoid tribe Ormiini, Tachininae). Some of the characters found in both Mesnilana and Rhinophoroides are the small and tongue-shaped lower calyptra that diverges from the scutellum and the parafacial with several setulae; these traits are also found in the coleopteran parasitoid tribe Palpostomatini. Besides these traits, the general appearance (abdominal chaetotaxy and head proportions) is very similar to some Palpostomatini, mainly the genus Palpostoma (e.g., Palpostoma subsessile Malloch, 1931). Based on these observations, Mesnilana and Rhinophoroides are removed from Dufouriini and tentatively considered as Palpostomatini, until additional evidence becomes available. In Stireman et al. (2019), Palpostomatini was a polyphyletic group, with one part forming a clade with Imitomyiini and sister to all other Dexiinae + Phasiinae, and another part as sister to Freraeini.

For a long time, Dufouriini was considered a tribe or subtribe of Phasiinae. It was initially allocated as a subtribe of Phasiini by Mesnil (1939), and then as tribe of Phasiinae by Emden (1945, 1950) based mainly on chaetotaxy. Verbeke (1962, 1963) considered it as a new subfamily: Dufouriinae, including two tribes, Dufouriini and Macquartiini (the latter currently in Tachininae), based mainly on postgonites of the intermediate type (in relation to the sensory and the connective Type II) and distiphallus DEG subtype. Verbeke also noted similarities in the male postabdomen shared by Dufouriinae and Phasiinae and was the first to suggest a close relationship between Dufouriinae and Phasiinae. Finally, the specializations of the female terminalia which allow Dufouriini to parasitize adult Coleoptera, as well as Phasiinae to parasitize adult Heteroptera, support the proximity between the two groups (Verbeke 1962). In contrast, in Dexiini hosts are actively sought out by first instar larvae deposited by females near the host and females possess a simple and short terminalia, with larvae completing their development in the host (Barraclough 1992). Later, Crosskey (1976, 1980) also recognized the subfamily Dufouriinae with the tribes Imitomyiini and Dufouriini, as these two would be excluded from Phasiinae and Dexiinae, respectively.

Following Herting (1984), Tschorsnig (1985) considered Dufouriini as a tribe of Dexiinae, with this subfamily as probably monophyletic, being supported by characters of the male terminalia; however, he recognized it as very inconsistent considering its biology and adult external characters. As discussed previously, the main putative synapomorphy discussed by Tschorsnig (1985) – aedeagus with basiphallus and distiphallus articulated to each other – was not recovered as a synapomorphic character herein, agreeing with Cerretti et al. (2014). The state 1 of character 130 is a synapomorphy shared by clade 1 (Dexiini + Voriini) and clade 4 ((Freraeini (Oestrophasiini + Dufouriini)) + Phasiinae) but undergoes a reversal in Phasiinae. Cerretti et al. (2014) proposed the paraphyly of Dufouriini s.l. in relation to Phasiinae, providing more evidence for a close phylogenetic relationship between these groups. Furthermore, Tschorsnig (1985) also recognized a number of similarities between the male terminalia of Dufouriini and Phasiinae, reporting that only the pregonite and phallus would place Dufouriini near Dexiinae. Considering his dichotomous key of the male terminalia of Tachinidae (Tschorsnig 1985), several shared characteristics can be found in the couplet of Dufouriini and Phasiinae: sternite 5 without lobes and without lateral membranous line; membranous connection between sternites 5 and 6; tergite 6 fused to segment 7 + 8. In the same line, Cantrell (1988: 147) stated: “The affinities of the Dufouriinae appear to be intermediate between those of the Phasiinae and Dexiinae and deserve further study.” Barraclough (1992) reported that the Palaearctic Dufouriini would not belong to Dexiinae, considering modifications in the female terminalia (elongated tergite 8 forming dorsal lamellae). He then affirmed: “[T]he Dufouriini belong in neither the Phasiinae nor Dexiinae.” (1992: 1152).

Our phylogenetic results support the proximity between the clade (Freraeini (Oestrophasiini + Dufouriini) and Phasiinae, as previously suggested by Verbeke (1962, 1963), Crosskey (1976, 1980) and Cerretti et al. (2014). Furthermore, Verbeke (1962, 1963), Crosskey (1976, 1980), Cantrell (1988) and Barraclough (1992) indeed argued for Dufouriini as a separate subfamily, i.e., Dufouriinae. Despite our results, other relevant phylogenetic results (Cerretti et al. 2014; Stireman et al. 2019) were not conclusive in supporting (or rejecting) the ideas of a clade formed by Dufouriini, Oestrophasiini and Freraeini or a close relationship between this clade and Phasiinae. Our taxonomic sampling, with five out of 12 Dexiine tribes (Dexiini, Voriini, Dufouriini, Freraeini and Oestrophasiini), does not allow any conclusions at the subfamily level. Hence, we included a comprehensive sampling for Dufouriini, Oestrophasiini and Freraeini, but a reduced and critical sampling of other Dexiine tribes. Cerretti et al. (2014) sampled five tribes (Dexiini, Dufouriini, Eutherini, Freraeini, Voriini) and Stireman et al. (2019) included representatives from all Dexiine tribes, and Dufouriini + Oestrophasiini was not closely related to Freraeini, nor was it close to Phasiinae. However, many tribes were not monophyletic (namely Dexiini, Voriini, Palpostomatini, and Dufouriini), perhaps indicating the need for more information (e.g., phylogenomic approaches and a detailed morphological analysis; and/or the need for better sampling of each tribe). This matter is completely open to debate with Dexiinae deserving further studies to reach a better conclusion about the systematic ranking and placement of Dufouriini, Oestrophasiini and Freraeini. Only time and more empirical data will tell whether these three tribes should be better elevated to subfamily level (the Dufouriinae of Verbeke 1962, 1963).

We propose a new classification for Dufouriini based on our phylogenetic results (see Supplementary file 4). The tribe Dufouriini is redefined and restricted now to five genera only: Chetoptilia, Comyops, Dufouria, Ebenia and Rondania. Comyopsis is proposed as a junior synonym of Ebenia, and Ebenia neofumata Santis and Nihei nom. nov. is transferred from Comyopsis to Ebenia. The other genera formerly recognized in Dufouriini are allocated to Freraeini and Oestrophasiini. The tribe Freraeini is redefined and broadened to include Microsoma, Eugymnopeza and Pandelleia, along with the type genus, Freraea. The tribe Oestrophasiini sensu Guimarães (1977) is revalidated, including four genera: Cenosoma, Jamacaria, Oestrophasia and Euoestrophasia, all removed from Dufouriini. Cenosomastat. rev., previously a subgenus of Oestrophasia is revalidated as genus. Finally, although not included in the phylogenetic analysis, Mesnilana and Rhinophoroides are removed from Dufouriini and are tentatively transferred to Palpostomatini.