Costal vein chaetotaxy, a neglected character source in Fanniidae and Muscidae (Diptera: Calyptratae)

. The present study examines whether the chaetotaxy of the costal vein in the calyptrate families Fanniidae and Muscidae deserves more attention in phylogenetic and taxonomic contexts. An overview of the macrotrichia and their arrangement on wing vein C is given. Special attention is given to the presence/absence of ventral and dorsal setulae on the costal sectors CS1 ‒ 3. This is described as one variable character ( A ) with nine states ( A0 ‒ A8 ). Specimens of both sexes (when possible) of each species belonging to 4 of a total of 5 fanniid genera and 115 of a total of 179 muscid genus-group taxa were examined and scored for character A . It was found that the presumed ancestral state of character A di ﬀ ers between the two families. It is further shown that the main transformational trend in character A in Muscidae has been bi-directional, leading either to the loss of ventral setulae or the gain of dorsal setulae. The utility of character A in the Fanniidae and Muscidae is many-sided and involves taxa ranging from species to family. It is concluded that character A and other aspects of costal chaetotaxy deserve more attention in morphology-based studies of calyptrate ﬂ ies. reveals basal split in Muscidae ﬁ rms as ancestral feeding and reveals an evolutionary correlation


Introduction
The costal vein C (Fig. 1A-B) constitutes the fi rst longitudinal vein and the anterior edge of the wing in Diptera. As documented by Hackman & Väisänen (1985), the chaetotaxy of vein C in Diptera embodies a suite of variable characters of potential value in taxonomy and phylogeny. Still, this chaetotaxy has generated little attention in most groups of Diptera, including the major radiation known as the Calyptratae. Only the spines at the subcostal break, or more extensively on vein C, are regularly mentioned in taxonomic texts in case these are notably enlarged.
Ongoing attempts to construct a practical key to the genera and species of the calyptrate family Anthomyiidae of NW Europe based primarily on external characters made me aware of the utitily of characters from the chaetotaxy of vein C. One variable character in particular, viz to what extent the ventral and dorsal surfaces of vein C are bare or setulose, turned out to be a valuable source of reliable states worth incorporating in identifi cation keys, diagnoses and descriptions. My experience with the Anthomyiidae made me curious to fi nd out whether the same character has been unjustly ignored in other calyptrate families. The present study attempts to answer this question in the calyptrate families Fanniidae and Muscidae.
The wing of calyptrate Diptera is as a rule covered on both sides in dense, hair-like microtrichia representing simple cuticular outgrowths (Fig. 1A). In addition, the wing veins are, to a varying extent, supplied with innervated macrotrichia of diff erent kinds, but clearly dominated by mechano-sensory sensilla in the shape of setulae, spinules and spines, depending on their length and strength ( Fig. 1A; se, sp). These are invariably present in dense rows on vein C and frequently also in small groups or loose rows on some of the other veins. Campaniform sensilla ( Fig. 1A; cs), detectors of cuticular stress, are found isolated or clustered on certain veins other than vein C in the basal part of the wing (Gnatzy et al. 1987). Instead, vein C has a loose dorsal row of fi ne, hair-like chemo-sensory macrotrichia (Figs 1A,2D;gs). These are no doubt homologous with the gustatory costal sensilla in Drosophila studied by Valmalette et al. (2015) and Raad et al. (2016).

Specimens examined
The results of the present paper are based primarily on the examination of specimens from the large collections of dry-mounted Fanniidae and Muscidae in the Natural History Museum of Denmark, Copenhagen, Denmark (NHMD), supplemented with species of Fanniidae from the Australian National Insect Collection, Canberra, Australia (ANIC).

Illustrations
Detached wings were mounted temporarily in 70% EtOH on glass slides with cover slips. The photos were taken with the use of a Leica MZ 16A microscope with a mounted Leica DFC420 camera. The software used was Leica Application Suite ver. 4, with built-in stacking facilities. Regarding the photos, wings pointing to the right are dorsal views, those pointing to the left are ventral views.

Costal chaetotaxy
Vein C in calyptrate fl ies (other than Hippoboscoidea, Scathophagidae and Oestridae) is nearly always armed with two parallel anterior rows of alternating spinules and setulae extending from the humeral break to well beyond the tip of vein R 4+5 (Fig. 1B). Both rows gradually diminish in strength towards the apex of the wing concomitant with the diminishing strength of vein C. This spinule-setula pattern, almost confi ned to calyptrate families, was denoted as costal chaetotaxy "Type C" by Hackman & Väisänen (1985). Frequently, the spinules at the subcostal break are to a varying extent enlarged and referred to as spines (e.g., Fig. 13A-B). In addition -and this is the main issue of the present study -vein C is bare or to a varying extent setulose on the ventral and dorsal surfaces behind the lower and upper spinulesetula rows. In order to describe this variation it is convenient to divide vein C into sectors determined by the costal breaks and terminations of veins Sc, R 1 , R 2+3 and R 4+5 in vein C. The costal sectors CS1-4 ( Fig. 1A) are presently delimited as follows: CS1 by the humeral and subcostal breaks CS2 by the subcostal break and the termination of vein R 1 in vein C CS3 by the terminations of veins R 1 and R 2+3 in vein C CS4 by the terminations of veins R 2+3 and R 4+5 in vein C The basal part of vein C, delimited by the costagial and humeral breaks, is not counted here as a costal sector because of its deviating, stable coverage of uniform longish setulae.
The observed variation in the development of setulae on the ventral and dorsal surfaces of vein C among Fanniidae and Muscidae is described below as a single multi-state character (A). This follows from my observation that dorsal costal setulae in fanniid and muscid species are only found in combination with a full coverage of ventral costal setulae.
Character A is divided into nine states as defi ned below. Notice that states A0-3 all agree with A4 in the absence of dorsal setulae on vein C and states A5-8 all agree with A4 in being extensively setulose ventrally on vein C.  A0 Vein C without ventral setulae on CS1-CS4, disregarding the occational presence of a few setulae at the extreme base of CS1 A1 Vein C with ventral setulae occupying less than the basal two-thirds of CS1 A2 Vein C with ventral setulae occupying all of CS1, sometimes with some irregular setulae on CS2, rarely even on the basal part of CS3 A3 Vein C with ventral setulae occupying CS2 and more than the basal half of CS3 A4 Vein C with ventral setulae occupying CS1, CS2 and more than the basal third of CS3; vein C bare or practically so dorsally on CS1-4 A5 Vein C with dorsal setulae primarily confi ned to the basal half or more of CS1, but sometimes with scattered setulae dispersed onto CS2 A6 Vein C with dorsal setulae primarily confi ned to CS2, but frequently with scattered setulae dispersed onto the adjacent costal sectors A7 Vein C with dorsal setulae on the basal half to two-thirds of CS3, but they sometimes reach onto CS2 A8 Vein C with dorsal setulae occupying CS1, CS2 and usually more than the basal half of CS3, sometimes even extending onto the basal part of CS4 A few species have the setulae from one or both anterior spinule-setula rows of vein C displaced ventrally or dorsally relative to the spinules. This is rarely seen in species of Fanniidae and Muscidae according to my observations, but a few cases are described below. Such displaced setulae obviously do not count as 'true' ventral or dorsal setulae. Likewise, the tiny and delicate gustatory sensilla ( Fig. 1A; gs) inserted in a loose row behind the upper anterior spinule-setula row should not be scored as dorsal setulae in the sense of character A.

Results
Class Insecta Linnaeus, 1758 Order Diptera Linnaeus, 1758 Section Schizophora Becher, 1882 Subsection Calyptratae Robineau-Desvoidy, 1830 Superfamily Muscoidea Latreille, 1802 Family Fanniidae Schnabl & Dziedzicki, 1911 A distinctive, modest-sized family known for ca 400 extant species currently classifi ed in fi ve genera (Table 1). Species belonging to four fanniid genera have presently been scored for the distribution of setulae dorsally and ventrally (character A) on vein C (Table 1) in agreement with the states described above.
In Piezura graminicola and P. pardalina (Fig. 2D) rows of ventral and dorsal costal setulae on CS1 represent setulae displaced from the anterior spinule-setula rows; these accordingly consist of spinules only.
This genus consists of Australofannia spiniclunis Pont, 1977 and one undescribed species, both endemic to Australia (Pont 1977). Examination of A. spiniclunis revealed a costal chaetotaxy deviating substantially from the usual fanniid pattern. Both sexes (Figs 3-4) have vein C extensively setulose both ventrally and dorsally (state A8). Further, extensive weakening of the costal spinules has taken place: the lower anterior costal row is wholly setulose in both sexes, whereas the upper anterior row has retained the usual spinule-setula pattern in the female (Figs 3A-C, 4A), but with only a few spinules distally on CS1 in the male sex (Fig. 4B).
Family Muscidae Latreille, 1802 The Muscidae with ca 5000 described species represent one of the largest radiations among the calyptrate families. The classifi cation adopted here is fairly conventional, morphology based, but slightly updated according to results obtained from analyses of molecular data presented in particular by Kutty et al. (2014). It recognizes eight subfamilies, seven tribes and 179 genus-group taxa (Table 1).
Species from 115 muscid genera and subgenera have presently been examined for the distribution of setulae ventrally and dorsally on vein C (character A) in agreement with the nine states described above (Table 1).   A subfamily of 'lower' Muscidae including the former Achanthipterinae Hennig, 1962(Kutty et al. 2014. Species from nine of a total of 12 recognized genera were examined (Table 1).
• Australophyra Malloch, 1923 The only known species, Australophyra rostrata (Robineau-Desvoidy, 1830), has the ventral costal setulae confi ned to CS1 (state A2). Ventral setulae are also seen on CS1 in Drymeia tetra (Meigen, 1826) and some D. alpicola (Rondani, 1871), but these are setulae displaced from the lower anterior spinule-setula row. Examination of a large number of species showed that vein C is bare dorsally and with a trend towards loss of ventral costal setulae. In Hydrotaea borussica, H. irritans and allied species, the lower anterior spinule-setula row on the basal part of CS1 consists of setulae only, some of which are frequently displaced ventrally.
Subfamily Muscinae Latreille, 1802 In this large, but well-defi ned and conceivably monophyletic subfamily, vein C is always bare dorsally, while the development of ventral setulae varies substantially among and within the genera.
Tribe Muscini Latreille, 1802 This tribe is now suspected of being paraphyletic, because molecular studies (Kutty et al. 2014(Kutty et al. , 2019 nest Stomoxyini within the Muscini as presently understood. Grzywacz et al. (2021), based on molecular approaches, even suggested a classifi cation in which the traditional Muscini genera Mesembrina and Polietes are moved to the Azeliinae. This classifi cation is not adopted here in the absence of morphological support. Species from 17 of 23 currently recognized genera and subgenera (Table 1) have been examined, but data from the remaining genus-group taxa (names marked with an asterisk below) have been extracted from Nihei & de Carvalho (2007), who performed a character analysis of the Muscini that involved scoring of the presence/absence of setulae ventrally on CS1 or CS1-CS3.  8C).
Three large species examined have the ventral costal setulae confi ned to CS1 (state A2): Polietes lardarius (Fabricius, 1781) (Fig. 10C) The specialized mouthparts of this group of hematophagous muscids leave no doubt as to their monophyly. Among the 10 currently recognized genera I have examined species belonging to fi ve of them (Table 1).

• Stygeromyia Austen, 1907
In the only examined species, Stygeromyia sanguinaria Austen, 1909, vein C is bare ventrally except for single setulae at the extreme base of CS1 (state A0).

Subfamily Atherigoninae Fan, 1965
This subfamily includes only one large genus of morphologically deviant Muscidae. It is obviously monophyletic, but its relationships remain uncertain.
• Atherigona Rondani, 1856 ( Fig. 11A-B) A total of 20+ named and unidentifi ed species of Atherigona from the Old World tropics and Mediterranean subregion were examined. They invariably share state A0, in which vein C is wholly bare ventrally behind the lower anterior row of alternating spinules and setulae, as exemplifi ed by A. orientalis Schiner, 1868 ( Fig. 11A-B).

Subfamily Reinwardtiinae Brauer & Bergenstamm, 1889
This taxon is a heterogeneous, para-or polyphyletic assemblage of genera which is also refl ected in its diverse costal chaetotaxy. Species from seven among the 15 currently recognized genera (Table 1) were examined.

Subfamily Phaeoniinae Malloch, 1917
This is a large and diverse, doubtfully natural subfamily divided in three tribes. It was found that all examined species invariably have vein C extensively setulose ventrally.
Phaonia angulicornis and P. steinii tend to have the setulae from the upper anterior spinule-setula row displaced dorsally on CS1.
Tribe Eginiini Stein, 1907[Syllegopterini Brauer & Bergenstamm 1889 Adult morphology and larval biology leave no doubt that this small tribe is monophyletic (Michelsen 2007). Species from two out of four recognized genera (Table 1) were examined.

Subfamily Coenosiinae Verrall, 1888
This large taxon is reasonably well defi ned and conceivably monophyletic (Couri & Pont 2000;Kutty et al. 2014). Among the two recognized tribes, Limnophorini and Coenosiini, only the latter is defi ned by alleged apomorphies. It was found that the great majority of the examined species representing 35 genera of Coenosiinae possess the prevalent muscid state A4, in which vein C is bare dorsally and extensively setulose ventrally. Species representing 12 out of a total of 24 currently recognized genera (Table 1) were examined.
Deviations from state A4 are unusual among limnophorine genera, but involve examples of both loss of ventral setulae and gain of dorsal setulae.
• Spilogona Schnabl, 1911 Examination of 40+ species of Spilogona revealed that the prevalent muscid state A4 is the rule in this large genus, but Michelsen (2021) described S. maderensis from Madeira (Portugal), which exceptionally is setulose dorsally on vein C (state A8).
• Tapantiomyia Michelsen, 2017 The only species, Tapantiomyia enigmatica Michelsen, 2017, is in the male sex (female unknown) unique among Coenosiinae (Michelsen 2017: fi g. 6) by the absence of ventral costal setulae, state A0, and further deviates from other Muscidae by the infl ated CS1, which is bare except for a single anterior row of tiny, pale setulae.

• Anaphalantus
The examined species of Neodexiopsis, Orchisia and Spathipheromyia have the setulae from the lower anterior spinule-setula row displaced dorsally, so that they occupy the usually bare strip separating the spinule-setula rows. The examined 50+ species of Coenosia all show the prevalent muscid state A4 in which vein C is bare dorsally and extensively setulose ventrally (Figs 1A-B, 18A-C). One exception was found in the boreal Palearctic species Coenosia emiliae Lukasheva, 1986 (Fig. 18D), which has dorsal costal setulae occupying CS2 and CS3 (state A7).

Phylogenetic and taxonomic implications
The monophyly of the present families Fanniidae and Muscidae is well supported by morphological (Chillcott 1961;Hennig 1965;Michelsen 2007;Domínguez & Roig-Juñent 2008) and molecular evidence (Kutty et al. 2014). Molecular studies attempting to illucidate relationships within the Calyptratae support that the major dichotomies took place in the following order (Kutty et al. 2008(Kutty et al. , 2019

Fanniidae
The costal chaetotaxy is remarkably uniform throughout the family Fanniidae. It consists of two parallel anterior rows of alternating spinules and setulae, as is the typical pattern in calyptrate fl ies other than Hippoboscoidea (Hackman & Väisänen 1985), but without additional ventral and dorsal setulae (state A0) other than the standard dorsal row of gustatory sensilla. Expectedly, the monobasic fanniid genus Zealandofannia described by Domínguez & Pont (2014) exhibits the same costal state.
Only Australofannia, known for two Australian endemic species, has a deviating and clearly derived costal chaetotaxy involving extensive replacement of the usual spinules by setulae and development of setulae ventrally and dorsally on vein C (state A8). A similar pattern is rarely seen in the major calyptrate radiation forming the sister group of Fanniidae, but has evidently been attained twice, in Scathophagidae and Oestridae. One other, less conspicuous deviation from the usual costal chaetotaxy in fanniids concerns the 'false' ventral and dorsal setulae on CS1 (Fig. 2D) here observed in two species of Piezura. This may prove autapomorphic for Piezura, a genus with only fi ve recognized species (Moores & Savage 2005).
There is no mention of the strange costal chaetotaxy of Australofannia in the description of the genus (Pont 1977) or in the morphology-based analysis of Fanniidae by Domínguez & Roig-Juñent (2008). It seems that the only focus on the costal chaetotaxy in the taxonomic literature on Fanniidae has been on the relative size of the costal spine at the subcostal break. Incidentally, this 'spine' consists of two subequal spines, one from each of the anterior spinule-setula rows. Normally, these spines are short and inconspicuous ( Fig. 2B-D), but quite strong in one species of Piezura as pointed out by Hennig (1955aHennig ( , 1955b and Chillcott (1961).

Muscidae
Two recent molecular analyses of the Muscidae by Kutty et al. (2014) and Haseyama et al. (2015) reveal major disagreements with respect to the basic muscid radiation and relationships. Both analyses challenge current morphology-based proposals on how to classify this comprehensive clade in subfamilies and tribes. It is expected that larger and more refi ned molecular analyses, combined with or corroborated by morphology, will eventually lead to more stability in the higher classifi cation of the Muscidae.
The present study revealed that the costal chaetotaxy in Muscidae is very diverse with respect to character A (development and distribution of ventral and dorsal setulae), as this varies through all states (A0-A8) described above. It also disclosed that state A4 (vein C extensively setulose ventrally and wholly bare dorsally) is very prevalent and found to be present in all muscid subfamilies and tribes apart from the small and uniform Atherigoninae and Eginiini (Table 1). This allows two assumptions: (1) A4 represents the ancestral state in Muscidae in contrast to the assumed ancestral state A0 in Fanniidae, and (2) the main transformation of character A during muscid radiation has been bi-directional, one leading to a loss of the ventral costal setulae (states A3→A0), the other leading to a gain of dorsal costal setulae (states A5→A8).
The dominant muscid state A4 may be present invariably in most genera of Muscidae including the specious Dichaetomyia, Myospila, Limnophora, Lispe and Pygophora. State A4 is also the rule in Thricops, Helina, Phaonia, Gymnodia, Hebecnema, Spilogona and Coenosia, but these genera contain one or more species with a deviating costal chaetotaxy Absence of ventral costal setulae is basic to the Atherigoninae, and a main trend towards reduction of ventral costal setulae is seen in the Azeliinae and Muscinae (Muscini, Stomoxyini). Exceptions to this are Huckettomyia and some Thricops (Azeliini) that have attained dorsal setulae on vein C.
Leaving aside the in many ways aberrant Atherigoninae, the above subdivision of the Muscidae determined by opposite transformations of character A coincides with the basal split of the family advocated by Kutty et al. (2014) on basis of molecular data.
A transformation of character A in terms of loss of ventral costal setulae is seen within the genera Hydrotaea, Musca, Neomyia, Polietes, Haematobia, Haematobosca, Stomoxys and Gymnodia. As would be expected, the most ancestral species of Hydrotaea, as judged from the larval mouth skeleton, number of instars and feeding habits (Skidmore 1985;Grzywacz 2013), are those without or with only a moderate loss of ventral costal setulae (states A4, A2). The opposite trend in terms of gain of dorsal costal can be observed in the genera Thricops, Helina, Phaonia, Hebecnema, Spilogona and Coenosia.
To the extent that the genera listed above are monophyletic, it seems safe to regard A4 as the ancestral state, except for the genus Musca starting out with state A2.
One case of character state reversal was found in Drymeia, a large genus characterized by the absence of ventral costal setulae (state A0). Only Drymeia hamata was found to have a ventrally setulose vein C (state A4), which is the prevailing and ancestral state in most muscid lineages but not in Drymeia, as proboscis morphology (Michelsen 2011) and molecular evidence (Savage & Sorokina 2021) place D. hamata in a subordinate position within Drymeia.
Males and females of the same species normally agree with respect to character A. Exceptionally, as seen in some species of Thricops, Helina and Phaonia, the dorsal costal setulae are more abundant or confi ned to the female sex or to some female specimens. In Pyrellia a stable diff erence determined by gender was found in the distribution of the ventral costal setulae. Within-species variation of character A not determined by gender is unusual, but was found in a few species of Thricops, Helina, Phaonia and Mydaea.
In the following are given some examples illustrating the practical use of the costal chaetotaxy, in particular character A, in systematic work on Muscidae.
Azelia and Thricops are the only genera with setulae at the inner hind margin of the hind coxae. These genera are easily separated by vein C, which is bare ventrally in Azelia but setulose ventrally in Thricops.
As discussed by Zielke (2016), Spilogona includes several densely setose species that on external facies resemble and have been confused with species of Drymeia, albeit these genera are distantly related. However, the present study revealed that vein C on the ventral side is setulose in Spilogona and bare in Drymeia. Only D. hamata, a species with strikingly long and slender labella (Michelsen 2011), agrees with Spilogona by having ventral costal setulae. Zielke (2016) described a second species of Drymeia with extended labella, but it remains to be determined whether it is bare or setulose on the ventral side of vein C.
The presence of dorsal costal setulae in Huckettomyia, a detail overlooked in the existing treatments of the genus (Pont & Shinonaga 1970;Savage & Wheeler 2004;Pont 2005;Pont & Vikhrev 2009), may help to distinguish females of this genus from habitually similar species of Hydrotaea and Neohydrotaea.
The Stomoxyini include at least three genera (Haematobia, Haematobosca and Stomoxys) in which diff erences in the distribution of ventral costal setulae (A4, A3, A1) may help to separate similar species. This character received no attention in the World revision of Stomoxyini by Zumpt (1973).
Species of Atherigona are readily distinguished from superfi cially similar species of Coenosiini by the absence of ventral setulae on vein C (state A0).
Major disagreements prevail in how to defi ne and delimit the Reinwardtiinae and Cyrtoneurininae in order to attain monophyly (Couri & de Carvalho 2003;Schuehli & de Carvalho 2005;Soares 2008;Savage 2009;Haseyama & de Carvalho 2011Patitucci et al. 2011;Pereira-Colavite & de Carvalho 2012;Kutty et al. 2014;Haseyama et al. 2015Haseyama et al. , 2019Pérez et al. 2020). Both subfamilies contain many small Neo-and Afrotropical genera with unsettled relationships. It was observed here that the Reinwardtiinae and Cyrtoneurininae as presently delimited both include genera that have vein C either bare or extensively setulose dorsally, an aspect that perhaps deserves more attention in future studies on these problematic groups.
The examined species of Thaumasiochaeta have dorsal setulae on CS1 (Fig. 17A-B). This costal chaetotaxy (state A5) is unique among the examined genera of Limnophorini.
Both examined species of the small Neotropical genus Cordiluroides have dorsal costal setulae on CS2 and CS3 (state A7). This is an unusual and derived state within the Coenosiinae (presently seen elsewhere in Coenosia emiliae) that may help to characterize and distinguish Cordiluroides from allied genera, provided it turns out to be stable at the species level.

Conclusions
There exist numerous morphology-based studies attempting to reveal phylogenetic relationships within the Fanniidae and Muscidae at family-group or genus level. These contain as a rule a list of variable, two-or multi-state characters considered relevant for scoring in a selection of species. My search through such character lists revealed that these often include characters from the wing chaetotaxy, but practically never from the chaetotaxy of the costal vein. Only an analysis of the Muscini by Nihei & de Carvalho (2007) includes a three-state character describing variation in the development of setulae on the ventral side of vein C, i.e., equivalent to character A. In a follow-up by Nihei & de Carvalho (2009) treating the genus-group taxa of Muscini in a key and by individual diagnoses the state of the ventral costal setulae is only mentioned for Mesembrina and Pyrellia.
It seems safe to conclude that the present comparative study of the costal chaetotaxy in Fanniidae and Muscidae has paid off by disclosing a wealth of previously unnoticed character data and patterns of immediate use in various systematic contexts.
Uluru macalpinei (Ulurumyiidae), a possible sister species of the remaining ca 15000 species of Oestroidea (Kutty et al. 2019), is extensively setulose on both sides of vein C (state A8) according to Michelsen & Pape (2017). That family aside, it seems that information about character A in the oestroid fl ies is fragmentary or missing for several groups. Observations made on the NW European fauna of blowfl ies (Calliphoridae s. lat.) by Rognes (1992)