﻿A revision of Centrioncus Speiser (Diptera, Diopsidae, Centrioncinae) with descriptions of new species from Angola, Burundi, and Kenya

﻿Abstract A diagnosis is presented for the Centrioncinae, the Afromontane Forest Flies or stalkless Diopsidae, while its taxonomic position within the Diopsidae is discussed. Arguments are presented for an eventual raising of the Centrioncinae to family level. The differential characters for its two genera, Centrioncus Speiser and Teloglabrus Feijen, are tabulated. The diagnosis for Centrioncus is updated and a key to the ten species now recognised (including three new species) is provided. Centrioncuscrassifemursp. nov. is described from a single female from Angola. This greatly extends the distribution range for the genus. Centrioncusbururiensissp. nov. is described from Burundi, while Centrioncuscopelandisp. nov. originates from the Kasigau Massif of Kenya. Diagnoses, descriptive updates, illustrations and notes are presented for all other Centrioncus. Centrioncusaberrans Feijen, described from Uganda, is now also recorded for western Kenya, Rwanda, and possibly eastern DR Congo. This wide range of C.aberrans is unusual for the Centrioncinae species which have allopatric and usually very restricted distribution ranges. Defining characters of C.aberrans from the various regions were examined in detail, but only minor differences were found. Centrioncusdecoronotus Feijen, described from Kenya, is now recorded for several other places in Kenya. A distribution map is given for the Eastern African Centrioncus species. The eastern branch of the Great Rift Valley appears to form a barrier between C.aberrans and C.decoronotus. The type species of the genus, C.prodiopsis Speiser from the Kilimanjaro in Tanzania, was only known from the 1905–1906 type series. After more than 100 years it has now been found again on the Kenya side of Kilimanjaro. Various differential characters of Centrioncus and Diopsidae are discussed, while brief discussions on sex ratio and fungal parasites are given. Centrioncus are known to occur on low shrubs and herbaceous plants in rain forests. Now, the possibility is indicated that they also might occur higher up in the tree canopies.


Introduction
It appears that the last word on the position of the Diopsidae within Diopsoidea or Acalyptratae has not yet been published. A rather different view on the position of Diopsidae and Syringogastridae was, for instance, given by Wiegmann et al. (2011) in a phylogenomic estimate of fly relationships especially based on molecules from 149 of 157 families. They placed the Diopsidae as the sister group of the Marginidae + Nannodastiidae + Canacidae, while the Syringogastridae were placed remarkably distant and as sister group of the Psilidae. These views were repeated by Wiegmann and Yeates (2017) in a review of phylogeny of Diptera. However, the results of Bayless et al. (2021) were different. Their results were based on analysis of protein encoding sequence data from transcriptomes, including 3145 genes from 70 species, representing all schizophoran superfamilies. They recovered Diopsidae as the sister group to the Syringogastridae, but beyond that, their relationships were unresolved. An overview of publications on the taxonomic position of the Diopsidae is given by Feijen and Feijen (2021). This latter paper also presents a diagnosis for the Diopsidae including the Centrioncinae.
McAlpine (1997) discussed several newly distinguished differential characters for the Centrioncinae, such as the supra-alar carina, pitting along the median ventral suture of the sternopleura, and the sawlines on various tarsal segments. In 2004, De Meyer described as new species Centrioncus bytebieri from the Taita hills in Kenya and discussed the stalkless Diopsidae. McAlpine (2011) described and illustrated the antenna of C. decoronotus.
The diagnoses for Centrioncinae and Centrioncus will now be updated. We will describe C. crassifemur sp. nov. from Angola. Furthermore, C. bururiensis sp. nov. will be described from Burundi and C. copelandi sp. nov. from Kenya. For Eastern Africa, new distribution records will be given for C. aberrans Feijen, C. bytebieri De Meyer, C. decoronotus Feijen and C. prodiopsis Speiser, while additional morphological and biometrical data for these species are presented. A distribution map is drawn for the seven Centrioncus species now known to occur in Eastern Africa. For the other Centrioncus, additional data are given, while photographic illustrations are presented for most species. A key to the ten Centrioncus species, now distinguished, will be given. The taxonomic position of the stalkless Diopsidae will be discussed. The relationship between Centrioncus and Teloglabrus will be considered as will be the intrageneric relationships in Centrioncus. Furthermore, various Centrioncus characters, sex ratio and fungal parasites will be discussed.

Materials and methods
Details on procedures for preparing genitalia slides and procedures for taking measurements are given in Feijen et al. (2018). For information on morphological terminology the reader is referred to the same source. For focus stacking photography of specimens, a Zeiss Stereomicroscope SteREO Discovery.V20 was used. Wings were photographed while mounted in slides. As useful differential character, the ratio width/length of the anterior sclerite of female sternite 7 is now added to the Centrioncus descriptions. For this ratio, the width is measured halfway along the length of the sclerite and the length is measured on the meson. The distribution map was built using the online version of SimpleMappr (Shorthouse 2010).
The morphological terminology for the abdomen of Centrioncinae broadly follows the one used by Feijen (1983). However, the sclerite(s) erroneously identified by Feijen (1983) as female sternite 8, were later identified as posterior sclerite(s) of sternite 7, while tergite 9 and sternite 9 in Feijen (1983) were subsequently referred to as tergite 8 and sternite 8 (see Feijen 1987). The aedeagus is now referred to as the phallus. For the femora, the indications outer side and inner side are used. Outer side stands for the side in lateral view (as in Fig. 117) and inner side for the reversed side. For legs, anterior and posterior side are also used (Cumming and Wood 2017), but that can create some confusion. For the fore femora, the outer side represents the posterior side and the inner side the anterior side, while for the mid and hind femora the outer side is the anterior side and the inner side the posterior side. For the bulbous posterior section of the pleura, the term pleurotergal dome is used. Likewise, for the spine on the posterior section of the pleura of stalk-eyed Diopsidae the term pleurotergal spine is used. However, Lonsdale (2020) places these structures more precisely on the katatergite, the ventral section of the pleurotergite. Feijen (1983) used as terminology for the male genitalia outer and inner surstyli (as telomeres) and interparameral sclerite. Lonsdale (2020) described the inner surstylus as "a lobe of the subepandrial sclerite". Feijen (1983) assumed the inner surstylus to be "probably of interparameral origin", so also as part of the subepandrial sclerite (sternite 10). The use of the terms "inner surstylus" or "medial surstylus" was discussed by Norrbom et al. (2019) and Sueyoshi (2005), who also considered it as a "lobe connected basally to the subepandrial sclerite and usually closely associated with the lateral surstylus". However, the use of the terms outer and inner surstyli for Centrioncinae is now discontinued. The outer surstylus becomes simply the surstylus. The inner surstylus is indicated as the clasper-like lobe of the subepandrial sclerite, for which as short form the term subepandrial clasper will be used. The interparameral sclerite is now indicated as subepandrial sclerite, while the mesal plate-like section will be referred to as subepandrial plate. The epandrial sclerite is not plate-like or even almost absent in stalk-eyed diopsids, but can also be present in the form of a slender rod, referred to as processus longi. The following institutional codens and abbreviations are used:
Diagnosis of Centrioncus. Updated version of the diagnosis by Feijen (1983). Centrioncinae with dark maxillary palps; dark section of funiculus limited to around base of arista; central region behind ocellar tubercle usually dark; wing with cell c partly or wholly glabrous, distal section of vein M4 gradually thinner, central wing spot present (Figs 1-8); on average 7-10 spinous setae per fore femur; basal ring of ♀ segment 7 with or without sutures; ♀ tergite 7 well sclerotised and with lateral edges curved under, ♀ sternite 7 split into a broad rectangular anterior plate and a curved or rectangular posterior plate (Figs 14,104,133,148); ♀ 7 th spiracles in tergite or membrane; subanal plate large; epandrium broad and rounded with ratio width/length 1.4-1.5; inner arm of surstylus quite detached from common base of outer and median arms; surstylus in dorsal to dorsolateral position (due to inversion appearing ventrally); outer and median arms with patches of microtrichia on outer side; connection between surstylus and subepandrial clasper short; subepandrial clasper glabrous, without ridges, with 1-3 long setae and 2-8 short setae; inner posterior corner of epandrium without mesad extension for articulation with subepandrial clasper; articulation between subepandrial clasper and cercus via 1-2 small mesad sclerites, the sclerites of both sides linked via a membranous connection; ♂ cercus with distally a broad lateral extension or cercus slender; subepandrial sclerite anteriorly with large lateral extensions, w/l ratio 2.0-2.5; epandrial fold broad and short or epandrial sclerite present; hypandrial clasper with three terminal setae or absent; epiphallus well denticulated, lateral sides of phallophore distally acuminate, phallus rather broad, distal phallic sclerites large and U-shaped; ejaculatory apodeme + sac very large (9.2-16.3% of body length), proximal section of ejaculatory duct at right angles to ejaculatory apodeme.
Remark. Centrioncus species are known from Angola, West Africa, and eastern Africa north of the Zambesi (north of 16°S latitude).

Key to the genera of Centrioncinae and species of Centrioncus
Although several Centrioncus and Teloglabrus have striking colour patterns of black and reddish brown on the thorax, this pattern is often not visible in preserved specimens, while it can sometimes also not be discerned in photographs of live flies. Wing patterns of central and apical wing spots are subtle and often cannot be distinguished in mounted flies or photographs of live flies. Mounting a detached wing in a slide is the best way to view the pattern. For confirmation of identifications, study of female and/ or male genitalia is required. A problem for the Centrioncus key is that only females are known for three species (C. angusticercus Feijen, C. decellei Feijen, and C. crassifemur sp. nov.), so corroborating male characters cannot be verified. The unique holotypes for C. angusticercus and C. decellei were not available for study, so newly used characters, such as ratios involving the scutellar spine and the state of sternites 4, 5, and 6, could not be verified. Given the usually restricted distribution ranges of the allopatric species, the location of origin forms an important indication of the species involved. Only C. aberrans and, to some extent C. decoronotus, are known to have a broader distribution range.
The first couplet of the key separates the genera Centrioncus and Teloglabrus. This issue is also treated in Table 1 where the major differences between these genera are listed. In addition, the differences are considered in the discussion section under "Distinctive characters for Centrioncus and Teloglabrus".
Supplementary description. Below, biometrical data are given for the much larger series now studied, as compared to the type series. Additional morphological data, as well as some rectifications, are presented.
Measurements. For the small type series of 1 ♀ and 2 ♂ Feijen (1983) gave for width of head in ♀ 1.12 mm and in ♂ 1.24 and 1.15 mm, body length in ♀ 4.9 mm and in ♂ 5.2 and 5.0 mm, wing length in ♀ 4.4 mm and in ♂ 4.7 and 4.4 mm, and length of scutellar spine in ♀ 0.26 mm and in ♂ 0.27 and 0.26 mm. In Table 2, measurements and other quantitative characters are presented for the much larger series now studied. In this table, data are presented for females and males separately. From this table, the differences between females and males for quantitative characters are marginal. The  body length is slightly larger in the male. In Tables 6, 7 the data for females and males are combined, so that large series can be compared with the other Centrioncus species for which large numbers were available. In the other three species for which large series could be measured, the females were, on the average, clearly somewhat larger. The original measurements of the type series of C. aberrans fall well within the ranges as presented in Table 2.
Colour. Most of the specimens now studied were not so dark as those of the type series. These latter three specimens were therefore probably somewhat discoloured, a common tendency in the Centrioncinae (Feijen 1983). The overall colour pattern can be seen in Figs 9-12.
Wing. Subcostal cell not visible in most specimens, visible in one specimen from Rwanda and one specimen of Mt. Elgon; vein CuA+CuP from vein CuP onward slightly curving downward under angle of 30° to wing margin (Fig. 1); central wing spot distinct, medium-sized, largely in basal quarter of cell r4+5, extending into cells br and bm+dm; vague infuscation along vein M4; cell cua in between triangular and rectangular ( Fig. 1).
Distribution and habitat. In the map of Eastern Africa (Fig. 29), the collection localities for C. aberrans are indicated. The eastern branch of the Great Rift Valley appears to form a barrier between the populations of C. aberrans and C. decoronotus. The two females collected in DR Congo/Rwanda are not indicated on the map. These two flies were collected by canopy fogging of individual trees. The methodology used for this fogging is explained in Wagner (2000Wagner ( , 2001. Funnel-shaped sheets were attached to the trees at waist height for collecting the dropping insects. Dr Wagner (pers. comm.) explained that "The lowest branches hit by the fog were about 3 metres [high] and it reaches usually ≤ 10 m." Feijen (1983) indicated the habitat for Centrioncus and Teloglabrus was low shrubs and herbaceous plants. We collected the flies at heights of ~ 20-60 cm from ground level. The results of the canopy fogging trials possibly indicate that Centrioncus flies might also occur at higher levels in the trees, though disturbance of the flies or descending insecticide might form alternative explanations. Nothing is known about where oviposition occurs and larval feeding, so the possibility of breeding in the tree canopy cannot be excluded. The flies of C. aberrans from Timboroa Forest in Kenya were collected in a  malaise trap. The height at which these flies were trapped is more in line with the habitat indicated by Feijen (1983). The Timboroa habitat is illustrated in Fig. 30.
Remarks. Given the unusually large range of distribution of this Centrioncus species, special care was taken to examine and compare defining characters such as surstylus, subepandrial clasper, male cercus and female sternite 7 for flies from the regions involved: western Uganda, western Rwanda, western and more central Kenya (Fig. 29). Comparisons of surstyli (Figs 21-24) and subepandrial claspers (Figs 25-28) lead to the conclusion that they are largely similar. For the surstyli, the general shape, number of tubercles and number of spinous setae correspond well. Also, the very specific female sternite 7 (Figs 13, 14) is very similar to the female paratype. The subepandrial clasper of the paratype (Fig. 25) appears less broad, but this is caused by it being slightly tilted to the right during the preparation. While comparing the shapes of the surstyli, care must be taken to present these three-dimensional structures as much as possible in the same plane. In the drawing of the surstylus of the paratype male (Fig. 21), the median arm is slightly tilted upward, so this arm appears somewhat shorter. Likewise, the inner arm of the surstylus in the Lac Gando specimen (Fig. 23) is somewhat tilted and so appears a bit different. In addition, small differences in shape of the three arms and size of setulae are observed and it could be argued that this species is in a very early phase of allopatric speciation. However, if one looks at the major differences for these defining characters between the Centrioncus species (Figs 51, 69, 88, 122, 138, 150;Feijen 1983), the conclusion can be drawn that all specimens examined across the large distribution range belong to one species, C. aberrans.
Diagnosis (after Feijen 1983). Centrioncus angusticercus can be recognised by its pruinose, mesally slightly depressed frons with two glossy spots; dark brown collar; blackish brown scutum; blackish brown scutellum with pale brown lateral sides and spines; pleura blackish brown except for brown propleuron and anterior anepisternum; apical seta/scutellar spine ratio: > 1.0; pale brown fore femur with distal third on inner side dark brown and small dark spot apically on outer side, with 36 tubercles; small round central wing spot (Fig. 31) around junction of crossvein r-m and vein M1 in tip of cell br and base of cell r4+5 slightly extending into cell bm+dm; tergites blackish brown, apical edges paler; female 7 th spiracle just in tergite; anterior sclerite of female sternite 7 rectangular, w/l ratio: ~ 4.4; posterior sclerite of female sternite 7 consisting of well-sclerotised trapezoidal anterior plate and two weakly sclerotised posterolateral extensions (Fig. 32), anterior edge of anterior plate parallel to its posterior edge, short    rior edge of anterior sclerite, short lateral sides straight, slightly expanded posteriorly, posterior edge concave; weakly sclerotised posterolateral extensions irregularly shaped and posteriorly with 3-4 pairs of setulae; very elongate female cercus with l/w ratio: 5.4 (Table 8).
Distribution and habitat. In the Shillito archive (now in NHMUK), a letter was found written on 22 January 1950 by D.J. Lewis, the collector of the holotype, and addressed to J.F. Shillito. It provided information on the collecting locality of the single known specimen: "The Diopsid, Centrioncus prodiopsis Speiser, was taken at an altitude of about 1980 metres among vegetation near a small stream in a patch of forest. Nagichot is on the Didinga Hills, one of the many isolated ranges, and may be considered as part of the Eastern and Southern zoogeographical Province ..." The type locality is shown on the map for Eastern Africa (Fig. 29).
Remarks. Feijen (1983) described the posterior sclerite of sternite 7 as "very roughly U-shaped with eight hairs, anterior edge parallel to posterior edge of sternum 7, anterior section well sclerotized, but arms of the U connecting to posterior hairs of sclerite weakly sclerotized." However, the view that this sclerite can be described as "U-shaped" is now rejected (compare Fig. 32 vs. Figs 43,62,82,119,133,148). Centrioncus angusticercus is now placed outside the group of Centrioncus with a U-shaped sclerite. Instead, it is considered as closer to the other Centrioncus without a U-shaped sclerite: C. aberrans and C. crassifemur sp. nov. Other similarities like l/w ratio of ♀ cercus and w/l ratio of ♀ anterior sternite 7 also support this group. For a confirmation of this relationship, the discovery of the male sex and study of the male genitalia of C. angusticercus and C. crassifemur sp. nov. and molecular analyses will be required.
Wing. Almost transparent but slightly tinged; small, central brownish spot around crossvein r-m in distal tip of cell br and basal quarter of cell r4+5, vaguely extending into cell bm+dm (Fig. 2); some vague infuscation around vein M4 proximally of crossvein dm-m; glabrous basal areas only include cells bc and c and basal third of cell br; crossvein h distinct; cell sc almost closed; vein CuA+CuP from vein CuP onward extending under angle of 30° to wing margin in almost straight line (Fig. 2); vein M4 continuing distal of crossvein dm-m very slightly turning downwards towards wing margin; cell cua triangular; alula distinct; crossvein bm-m hardly indicated.
Egg. ♀ paratype contained 45 developing and developed eggs in abdomen. The eggs (Figs 47, 48) measured ≤ 1.10 mm in length, with slightly elevated longitudinal ridges spanning from anterior pole to posterior pole with fine, nearly hexagonal microstructure between ridges (Fig. 48).
Distribution and habitat. The new species is only known from the Bururi National Forest in Burundi. It was collected at an altitude of 1955 m. This small Afromontane Forest is relatively isolated from other similar forests. The type locality is shown on the map for Eastern Africa (Fig. 29).
Etymology. The specific epithet of C. bururiensis sp. nov. refers to the place of origin for the holotype, the Bururi Forest in Burundi.
Remarks. While cursorily examining the three flies of the Bururi Forest, the male fly appeared to be Centrioncus aberrans, given the apparently similar shapes of surstyli and cerci. However, the females with their clearly visible curved posterior sclerite of sternite 7 (Figs 42, 43) did not belong to the C. aberrans species-group. This led to the initial assumption that this might be the first case of two Centrioncus species occurring sympatrically. However, closer examination of the male genitalia revealed distinct differences with those of C. aberrans. The female is also different from other described Centrioncus with the characteristic sternites 4, 5, and 7 and the spermathecae. The subanal plate of C. bururiensis sp. nov. is more like those of C. decoronotus and C. prodiopsis, not like C. aberrans (compare Fig. 46 vs. Feijen 1983: figs 49-51). The w/l ratio of the anterior female sternite 7 is like the ones for C. decellei, C. bytebieri and C. prodiopsis, certainly not like the one for C. aberrans (Table 8), while the shape of posterior sclerite of sternite 7 is more like the one for C. decoronotus and certainly not like C. prodiopsis or C. bytebieri (compare Figs 119, 148, 62 vs. De Meyer 2004: fig. 6).
It remains puzzling that the male genitalia of Centrioncus bururiensis sp. nov. indicates a possible relationship with C. aberrans, while the female genitalia contradicts such a relationship. Analyses, including molecular analysis, of a series of fresh material are required to resolve this issue. The number of 45 eggs found in a paratype was quite high. Feijen (1983) found in various Centrioncus and Teloglabrus species > 20 developing and developed eggs and stated that a fecundity of between 25 and 50 eggs seemed likely. The gravid female paratype was collected in September, coinciding with the start of the short rains in Burundi.  x-13.xi.2011, 2 ♀ 24.i-7.ii-2012, 1 ♀ 8-22.ii.2012, 1 ♂ 19.iv-3.v.2012, 1 ♀ 27.v-10.vi.2012. In total 10 ♀ and 6 ♂ were studied.
Supplementary description. Below, biometrical data are presented for the series now studied, and compared to the type series. Additional morphological data, as well as a few rectifications, are presented.
Measurements. For the type series of 26 ♀ and 26 ♂, De Meyer (2004) stated body length 5.59 mm (4.95-6.80) and wing length 4.83 mm (4.55-5.10). In Table 3, measurements and other quantitative characters are presented for the series now studied. In this table, data are presented for females and males separately. The table shows that differences between females and males for quantitative characters are small. The body length and various other measurements are slightly larger in females. In Table 6, the data for females and males are combined, so that large series can be compared with other Centrioncus species for which large numbers were available. The original measurements of the type series agree well with the means and ranges as presented in the tables: body length 5.55 mm ± 0.09 vs. 5.59 mm and wing length 5.04 mm ± 0.06 vs. 4.83 mm. The measurements confirm C. bytebieri as one of the larger Centrioncus, together with possibly C. angusticercus and C. bururiensis sp. nov. (based on measurements presented in this paper (Tables 6, 7), and Feijen (1983)). Colour. De Meyer (2004) gave a fine description of the colouration. Characteristic colouration of pleura in prime specimens should be stressed (Figs 55,57). Yellowishbrown sections include proepisternum, proepimeron, anepimeron with greater ampulla, posterodorsal corner of katepisternum and anterior spot on meron. Head. Frons (Fig. 56) thinly pruinose, laterally densely pruinose, with glossy spots (reduced in some specimens) laterally of ocellar tubercle; length of outer vertical seta 0.38 mm ± 0.01 (n = 16), length of fronto-orbital seta 0.28 mm ± 0.00 (n = 16) (see Tables 6, 7).
Wing. Central brownish spot around crossvein r-m in distal tip of cell br and basal fifth of cell r4+5, somewhat extending into cell bm+dm (Figs 3, 55); some infuscation around vein M4 proximal to crossvein dm-m; vein CuA+CuP from vein CuP onward extending under angle of 30° to wing margin in almost straight line (Fig. 3); cell cua triangular; vein M4 continuing distal of crossvein dm-m, gradually thinning, very slightly turned downwards towards wing margin.
Egg. Female from Ngangao carried nine developed and developing eggs in abdomen; eggs (Figs 66, 67) ≤ 1.04 mm in length; slightly elevated longitudinal ridges spanning from anterior pole to posterior pole with fine, nearly hexagonal microstructure between ridges (Fig. 67).
Distribution and habitat. The localities of the type series and the specimens now studied are indicated on the map for Eastern Africa (Fig. 29). The distribution appears to be limited to the Taita Hills, a Precambrian mountain range located in the Taita-Taveta County in south-eastern Kenya. These hills consist of three massifs: Dabida (Dawida), Sagalla and Kasigau. Centrioncus bytebieri is only known from the Dabida Massif. No Centrioncus are known from Sagalla, while C. copelandi sp. nov. occurs in Kasigau. The type series was collected in the forest remnants of the Dabida complex in the Ngangao, Chawia and Yale Forests, as well as in the Mbololo Forest. The new specimens also came from Ngangao, Chawia and Mbololo Forests, while as new locations the Fururu and Vuria Forests can now be added. De Meyer (2004) found that his collecting sites were situated from 1400-1800 m, while the specimens now studied were collected from 1680-2162 m. De Meyer added that the species is common and "found on leaves of lower vegetation (shrubs, plants), usually in shaded places in the forests." The gravid female was collected in the last week of January/first week of February, just before the start of the long rains in Kenya. The male collected in the last week of July/ first week of August proved to be teneral, so emerging after the end of the long rains.
Remarks. The additional specimens were largely from the type locality. Male and female genitalia, but also external characters like colour pattern and large size, conform to the description by De Meyer (2004). However, some differences and additions should be indicated. For the pleura, the brown anepimeron and greater ampulla also form part of the large central brown area. The wing has a central wing spot. De Meyer's description of the posterior sclerite of female sternite 7 was fine but the drawing ( fig. 6) shows a rather small sclerite. De Meyer gave a very brief description of the subanal plate. As this plate is very characteristic, it needed redescription. De Meyer described the female cercus as "not so narrow" but gave as l/w ratio: 4.2, which differs from the ratio 2.3 found here. Additional or more detailed illustrations have now been provided for various characters. Diagnosis. Centrioncus copelandi sp. nov. can be recognised by a mesally slightly depressed, pruinose frons with glossy spots; glossy collar; pruinose blackish brown scutum; dark brown scutellum with brown edges and spines; pleura blackish brown (Figs 74, 76); scutellar spine/scutellum ratio: 0.87-1.00; apical seta/scutellar spine ratio: 0.86-0.92; brown, strongly incrassate fore femur (l/w ratio: ~ 2.65) with ~ 31.7 tubercles, dark brown stripe dorsally on apical third of inner side; very large, distinct central wing spot in distal third of cell br, basal two-fifths of cell r4+5 and distal twothirds of cell bm+dm (Fig. 4); tergites blackish brown; female 7 th spiracle in tergite; sternite 4 rectangular; sternite 5 trapezoidal, invaginated anteriorly; sternites 4 and 5 with pair of strongly sclerotised spots; sternite 6 trapezoidal, 1.5× as broad as sternites 1-5; anterior sclerite of female sternite 7 with w/l ratio: ~ 3.1 (Fig. 81); posterior sclerite of female sternite 7 truncated U-shaped, straight anterior and lateral edges with angular anterolateral corners; female cercus elongate, l/w ratio: ~ 4.4; subanal plate pentagonal, apically tapering into short extension; spermathecae rounded, basally flat (Fig. 85), with small apical dimple; spermathecal ducts with constriction near spermathecae; outer and median arms of surstylus well separated, with short broad common base; outer arm constricted at base, apically broadening to 3× width of base, sides concave, with 16 apical tubercles (Fig. 88); median arm a slender rod, apically some long setulae but no spinous setae, slightly shorter than outer arm; inner arm twice as broad as median arm and slightly longer, apically with shape of cap opener; subepandrial clasper elongate, basal third constricted, apical corners angular, apically strongly convex; cercus (Fig. 89) with broad lateral extension on distal third, apically convex.
Wing. Almost transparent with very large, distinct, brownish central spot in distal third of cell br, basal two-fifths of cell r4+5, distal two-thirds of cell bm+dm and slightly extending into cells r2+3, m1 and m4 (Fig. 4); glabrous basal areas only include cells bc and c and small basal and subbasal spots in cell br; crossvein h distinct; cell sc closed; vein CuA+CuP from vein CuP onward extending under angle of 25° to wing margin in almost straight line (Fig. 4); cell cua triangular; vein M4 continuing distal of crossvein dm-m in almost straight line to wing margin; alula distinct; crossvein bm-m very vaguely indicated.
Male postabdomen. Tergite 6, sternite 8 and epandrium uniformly blackish brown; epandrium broad and rounded (too damaged to measure w/l ratio), clothed in microtrichia; outer and median arms of surstylus well separated, with short broad common base (Fig. 88); outer arm deeply constricted at base with concave lateral sides, broadening apically to ~ 3× basal width, apically with row of 16 tubercles, basally with nine setulae (2 very long) on inner side and a few small setulae along apical edge, on outer side with a few small setulae and central patch of microtrichia; median arm ( Fig. 88) slender, parallel-sided, rod-shaped, almost as long as outer arm, apically with four small setulae and two long setulae, no spinous setae present; inner arm twice as broad as median arm and slightly longer, apical quarter abruptly narrowed, cap opener-shaped, apically with seven setulae; subepandrial clasper ( Fig. 90) somewhat elongate, basal third constricted, apical corners angular, apically strongly convex, glabrous, with four long setulae on inner side; cercus (Fig. 89, Table 8) slender, with broad lateral extension on distal third, apically convex, clothed in microtrichia and short setulae, 7 longer setulae along apical edge, length/greatest width ratio: 1.5 (Table 8); ejaculatory apodeme + sac (Fig. 91, Table 9) very large, 10.1% of body length, ejaculatory apodeme slender, apically broadening.
Egg. Female with two damaged full-grown eggs and some pieces of undeveloped eggs in abdomen. Eggs (Fig. 86) measured 1.01 mm in length with elevated longitudinal ridges spanning from anterior pole to posterior pole with fine, roughly hexagonal microstructures between ridges (Fig. 87).
Distribution and habitat. The localities of the type series are indicated on the map for Eastern Africa (Fig. 29). The collecting sites varied from 1056 to 1117 m in altitude. The habitat of the Kasigau site at 1117 m is shown in Fig. 92. The distribution appears to be limited to the Kasigau Massif of the Taita Hills (see also under C. bytebieri). Centrioncus bytebieri is only known from the Dabida Massif of the Taita hills. The gravid female was collected in early July, in the middle of the dry season which is unusual for Centrioncus flies.
Etymology. The specific epithet of C. copelandi sp. nov. refers to the name of its collector Dr Robert S. Copeland (ICIPE), who contributed a significant number of Centrioncus specimens to this study.
Remarks. From the shape of the outer arm of the surstylus (apically much wider than at base, with large number of 16 or 17 tubercles), absence of spinous setae on the median arm of the surstylus, the very large central wing spot and anteriorly strongly constricted sternite 5, it can be postulated that the closest known relative of C. copelandi sp. nov. is C. prodiopsis. From a geographical point of view that also makes some sense (Fig. 29), although C. bytebieri occurs closer to C. copelandi sp. nov.
Egg. Holotype with six almost fully developed eggs in abdomen. Eggs (Figs 108, 109) measured 0.9 mm in length, with longitudinal ridges spanning from anterior pole to posterior pole with fine, nearly hexagonal, microstructures on and in between ridges. With different illumination settings, either ridges become more distinct (Fig. 109) or hexagonal microstructures (Fig. 108).
Distribution and habitat. The single known specimen is from north-western Angola. The holotype was most likely collected in rain forest around the experimental station of the Instituto de Investigação Agronomica de Angola at Quilombo. The altitude of the collecting locality must then have been between 600 and slightly > 1000 m. The gravid holotype was collected in early March, which falls in the second half of the rainy season in Angola.
Etymology. The specific epithet of C. crassifemur sp. nov. refers to the remarkably incrassate fore femora.
Remarks. The more sclerotised anterior margin of sternite 2 forms an integral part of sternite 2 (Fig. 101) and is not, like in other Centrioncus, with thin lateral connections tied to the main region of sternite 2. Given the shape of the posterior sclerites of female sternite 7, the closest known relative of C. crassifemur sp. nov. could be C. aberrans. However, it is possible that the shape of sternite 7 simply represents a symplesiomorphic condition. Centrioncus aberrans and C. bururiensis sp. nov. are geographically speaking its closest known relatives.
Supplementary description. Wing. Small, elongate, very vague, brownish central spot around crossvein r-m (Fig. 110), in tip of cell br, mainly in posterior part of basal quarter of cell r4+5 to three-quarters distance to crossvein dm-m, slightly extending into cell bm+dm; vein CuA+CuP from vein CuP onward extending under angle of 25° to wing margin in straight line; vein M4, especially distal of crossvein dm-m, curving downward to wing margin (Fig. 110).

Centrioncus decoronotus Feijen
Supplementary description. The biometrical data are presented for the series now studied, and compared to the type series. Additional morphological data as well as a few rectifications are presented. Various aspects of the morphology are now illustrated by photographs, while line drawings are presented for flies from Ol Doinyo Orok, a location distant from the type location.
Measurements. For type series of 3 ♀ and 4 ♂, Feijen (1983) provided the following measurements: body length ♀ 5.3 mm and ♂ 5.2 mm, head width ♀ 1.23 mm and ♂ 1.17 mm, wing length ♀ 5.0 mm and ♂ 4.7 mm, scutellar spine length ♀ 0.36 mm and ♂ 0.31 mm. For the much larger series now available, 22 ♀ and 22 ♂ were measured in more detail. In Table 4, measurements and other quantitative characters are presented for this series. In this table, data are presented for females well within the ranges provided here: body length 5.15 mm ± 0.04 (range 4.64-5.61, n = 44), head width 1.14 mm ± 0.01 (range 1.02-1.24, n = 44), wing length 4.51 mm ± 0.03 (range 4.09-5.00, n = 39), scutellar spine length 0.33 mm ± 0.00 (range 0.29-0.37, n = 44). Colour. The specific epithet decoronotus refers to the colourful notum of the mesothorax with its pattern of brown and blackish brown. This pattern is shown well in the fly from the Thomson's Falls (Fig. 112), but is much less pronounced to almost absent in flies from Katura Forest. This is not only due to preservation techniques, but is present in live specimens as can be observed in photographs of live flies from Ngare Naro forest (near Thomson's Falls) where the pattern is also less pronounced (Fig. 117).
Egg. Female from Ol Doinyo Orok forest with two almost fully developed eggs in abdomen. Eggs measured respectively 0.99 mm and 1.00 mm in length with longitudinal ridges spanning from anterior pole to posterior pole; in addition, more simple "lines". Ridges (indicated with R) with tiny elongate pits along their length, while lines (indicated with L) form more integral part of roughly hexagonal microstructures between ridges (Fig. 121).
Distribution and habitat. The collecting localities are shown on the map for Eastern Africa (Fig. 29). Centrioncus decoronotus was now found at altitudes varying from 1455-2329 m. Feijen (1983) reported it as occurring between 1200-2350 m. The eastern branch of the Great Rift Valley appears to form a barrier between C. decoronotus and C. aberrans. The gravid female was collected in March at the beginning of the long rains.
Supplementary description. Biometrical data are given for the remeasured series. Additional morphological data and illustrations are also presented.
Measurements. Feijen (1983) measured 20 ♀and 26 ♂ and gave as results: body length ♀ 5.61 mm ± SE 0.07 (range 5.2-6.2) and ♂ 5.28 mm ± 0.04 (4.8-5.7), head  Of the original type series, 10 ♀ and 10 ♂ were remeasured in detail. In Table 5, the new measurements and other quantitative characters are presented. In this table, data are presented for females and males separately. The table shows that differences between females and males for quantitative characters are small. Body length and various other measurements are slightly larger for the females.
The original measurements of the type series do not agree well with the measurements now recorded. The measurements now found for ♀ and ♂ combined are: body length 4.97 mm ± 0.06 (range 4.58-5.43, n = 20), head width 1.11 mm ± 0.01 (range 1.04-1.21, n = 20), wing length 4.28 mm ± 0.05 (range 3.78-4.64, n = 20), scutellar spine length 0.27 mm ± 0.00 (range 0. 24-0.29, n = 20). This means that the present measurements for body length and wing length are ~ 10% less, while the measurements for head width and scutellar spine are ~ 5% less. The specimens of the type series were measured in Malawi while in fresh condition. This could only affect the body length measurements to a small extent but not the other measurements. The persistent differences are probably due to calibration errors. The new measurements now indicate C. jacobae as a species is clearly smaller than the other Centrioncus species for which large series could be measured. (Tables 6, 7).
In connection with ecological research in Malawi from 1971 to 1975, the weight of fresh flies was regularly determined. The weight of some C. jacobae flies was also determined. In August (in the non-reproductive phase) seven females weighed on average 4.2 mg, while six males weighed on average 3.4 mg.
Colour. In optimal conditions, this species presents a colourful pattern of blackish brown and chestnut brown on the thorax as described by Feijen (1983) and now illustrated (Figs 126, 127).
Wing. Pale brownish with distinct, large central wing spot, covering more than basal third of cell r4+5 (just past crossvein dm-m), slightly extending into apex of cell br and extending into anterior section of cell bm+dm between crossveins (Fig. 7); distinct infuscation along crossvein dm-m and along vein M4 between cell cua and crossvein dm-m; vein CuA+CuP from vein CuP onward extending under angle of 30° to wing margin in almost straight line (very sightly curving downward); vein M4 continuing distal of crossvein dm-m in almost straight line to wing margin; cell cua triangular (Fig. 7).
Distribution and habitat. This species occurs at altitudes between 1300 and 1400 m in rain forests on the mountains around Blantyre and Limbe in Malawi. The rain forest at Mount Ndirande is gone, but patches on Mount Soche remain. The largest patches of rain forest remain on Mount Chiradzulu. However, this latter remark is quite relative when Loveridge's (1954) observations about the Chiradzulu forest are considered: "Unfortunately the forest was but a pitiful remnant of its former self, few of its trees exceeding thirty feet in height and even these patches of secondary growth were separated by bracken which had swept up from below" and "This deforestation has resulted in the drying up of many of the streams during much of the year". The strong discontinuity of the habitat of Afromontane Forest Flies will obviously be reinforced by the ongoing deforestation.

Centrioncus prodiopsis Speiser
Supplementary description. Feijen (1983) studied the ♂ lectotype and one ♀ paralectotype. The redescription by Hennig (in litt.) of the specimen of the type series in ZMHB also became available. Now two additional females from Kenya could be examined, but one of these two flies was rather teneral. Updated biometrical data are now given. Additional morphological data and illustrations are presented.
Head. As in lectotype and paralectotype, Kenya specimens with frons slightly depressed mesally, anterior quarter of pruinose frons less dark brown and on either side of ocellar tubercle with glossy spot (Fig. 141); length of outer vertical seta 0.31 mm (n = 2), length of fronto-orbital seta 0.22 mm (n = 2).
Wing. Almost transparent with very large, distinct, brownish central spot covering nearly basal half of cell r4+5 (well past crossvein dm-m), extending into distal half of cell br and extending into distal two-thirds of cell bm+dm and well into cells r2+3, m1 and m4 (Fig. 8); vein CuA+CuP from vein CuP onward extending under angle of 30° to wing margin in almost straight line; vein M4 continuing distal of crossvein dm-m in almost straight line to wing margin; cell cua subtriangular (Fig. 8).
Male postabdomen. Outer and median arms of surstylus well separated, with short, broad common base (Fig. 150); outer arm somewhat trapezoidal with straight lateral sides, broadening apically to almost twice width at base, apically with row of 17 tubercles with tiny setulae in between, with rectangular, raised section with 11 setulae basally on inner side (Fig. 150), outer arm wholly clothed with microtrichia on outer side; median arm slender, parallel-sided, slightly curved rod-shaped, slightly shorter than outer arm, apically with 4 small setulae and 2 long setulae, no spinous setae present, basal half of outer side covered with microtrichia; inner arm of surstylus short, ca. half-length of other two arms, with a bulbous preapical apophysis, apically with five setulae (Fig. 150); subepandrial clasper ( Fig. 151) elongate, slightly constricted at base, narrow, somewhat rectangular, apically with rounded corners, apical edge slightly convex, glabrous, with three long setulae centrally on inner side and four small hairs at apical edge; cercus basally narrow, broadening apically, with short, broad lateral extension, length/greatest width ratio: 1.4 (Table 8); ejaculatory apodeme + sac very large, 12% of body length (Table 9).
Distribution and habitat. The collecting localities are shown on the map for Eastern Africa (Fig. 29). The type locality is on the southern side of the Kilimanjaro region, while the Kenyan specimens are from the northern side of the Kilimanjaro region. In a straight line, the distance between the two localities is 55 km.
Remarks. Speiser (1910) based his description on two pairs of specimens. Feijen (1983) used for his redescription one pair from NHRS. In the NHRS collection, there was one additional fly without an abdomen, while ZMHB also housed one fly without an abdomen. Feijen designated the ♂ of the pair studied as lectotype and the ♀ as paralectotype. Now, a typed letter from W. Hennig, dated 3.xii.1947, was recovered from the J.F. Shillito archive (NHMUK). The section relevant to Centrioncus ran as follows: Ueber den Holotypus von Centrioncus prodiopsis Speiser, der sich in der Sammlung des Zoologischen Museums Berlin befindet, habe ich mir früher einmal, für den Fall, dass er zerstört werden sollte, Notizen gemacht. Kopien meiner Zeichnungen fuge ich für Sie diesem Briefe bei. Sie können diese Zeichnungen beliebig verwenden. Auch gegen Veröffentlichung habe ich nichts einzuwenden, da wir hier doch in absehbare Zeit nicht dazu kommen.
[In a brief translation this comes to: Regarding the holotype of Centrioncus prodiopsis in ZMHB, I made years ago notes in case it would be destroyed. I enclose copies of the drawings I made. If you like, you can use these for publication. I have not yet formed a conclusive judgment on the systematic position of Centrioncus prodiopsis. I only know for sure that the species does not belong to the Sepsidae or Megamerinidae. In view of the morphology of the copulation apparatus, there would hardly be anything to object to your view that it is a diopsid. However, I would like to reserve my judgment for the time being. On the Centrioncus head 1 vertical seta and 1 frontoorbital seta are present, while the thorax counts 1 notopleural seta (the posterior one), 1 supra-alar seta and 1 postalar seta.] From this letter, it became obvious that the fly without abdomen in ZMHB is a male, so the fly without abdomen in NHRS has to be a female. Hennig indicated the ZMHB fly as the holotype. However, Speiser did not designate a holotype. All four specimens of the type series were labelled as syntypes (Feijen 1983). It seems likely that Hennig was not aware of the three NHRS flies. As such, the designation of the lectotype by Feijen (1983) remains valid. Hennig's drawings of the ZMHB fly showed that this male was definitely conspecific with the lectotype. As, since the type series of 1905-1906, male specimens of C. prodiopsis have apparently never been collected again, it is considered useful to copy Hennig's drawing of the male postabdomen (Fig. 152) as an addition to Feijen's (1983) drawings and for historical reasons. In this drawing, Hennig's original labels are indicated. The ZMHB specimen is now designated as male paralectotype. Lonsdale (2013), in a review of Tanypezidae and Strongylophthalmyiidae, includes for the phylogenetic analysis in the character matrix as out-group representatives for the Diopsidae Sphyracephala subbifasciata Fitch and Centrioncus prodiopsis. However, the origin of the single Centrioncus studied was not given, though for the genitalia character states, Feijen (1983) was used. The two specimens from Kenya were, after 106 years, the first C. prodiopsis collected since the type series. Literature references to additional C. prodiopsis records related to other Centrioncus species or to Teloglabrus species (Feijen 1983).

Taxonomic position of the stalkless Diopsidae
Whether it will be necessary to keep Centrioncidae as a separate family remains to be seen. There are obviously striking differences compared to the stalk-eyed Diopsidae: no eye stalks, pubescent arista, ventrally extended funiculus, lanceolate basiliform proster-num, absence of pleurotergal spines, vein CuA+CuP reaching wing margin, presence of tubercles on hind femora, very large inverted male sternite 8 on both sides fused to sternite 7 which forms a complete ventral band of sclerotisation, and very different male genitalia (trilobed surstylus, presence of subepandrial clasper, small phallapodeme with posterior two-thirds fused to hypandrium and solid phallus with a complex distal section versus an open, delicate structure of sclerites and stylets coming together on a basal ring). On the other hand, Lonsdale (2020) lists as synapomorphies of the Diopsidae (including Centrioncinae) a scutellum with long apical spines, a bulbous katatergite, tarsi with dark "sawlines" lengthwise along the tarsomeres, absence of bmm, a porrect antenna, raptorial fore legs, and only one outer vertical and one frontoorbital seta. As far as DNA analyses is concerned, it appears that Centrioncinae and the stalk-eyed Diopsidae must have separated between 43 and 63 million years ago (F. A. A. Feijen, pers. comm. (2022); in Feijen et al. (2018) an introduction was given to her molecular analyses, but a full treatment is pending).
At present, we distinguish three subfamilies in the Diopsidae: Centrioncinae, Sphyracephalinae and Diopsinae (i.e., Diopsinae s.s.). This system was already proposed by Shillito (1971) and used by Feijen (1989) and Feijen and Feijen (2021). Hennig (1965) subdivided the Diopsidae into two subfamilies, Centrioncinae and Diopsinae (i.e., Diopsinae s.l.), while subdividing the latter subfamily into Sphyracephalini and Diopsini. Hennig's system was followed by Steyskal (1972) and Meier and Baker (2002). Hennig (1965) regarded his subdivisions as relative in nature and thought it possible that Centrioncinae and Diopsinae s.l. could be regarded as families or tribes. For an absolute allocation of category in the Acalyptratae, Hennig thought it best to consider subfamilies the narrowest monophyletic groups originating from before the appearance of Baltic Amber (Eocene). From the presence of Prosphyracephala succini (Loew) in Baltic Amber, Hennig concluded that the sister-group relationship between Centrioncinae and Diopsinae s.l. originated from before this period. Based on what was then (1965) known about fossil diopsids, Hennig considered it as quite likely ("durchaus wahrscheinlich") that in the Baltic Amber period only two diopsids were present: P. succini and a closely related ("sehr ähnlich") species from which the Diopsini originated. Based on several plesiomorphic characters in the Diopsini genus Diopsina Curran which are not present in Prosphyracephala Hennig, Feijen (1981) considered that both Sphyracephalini and Diopsini originated from before the Eocene and consequently should be given subfamily rank. Hennig's Centrioncinae and Diopsinae s.l. should then be elevated to family level. Hennig's (1965) assumption of only two diopsids present in the Baltic Amber period has now become dated. Lewis (1971) described the compression fossil Prosphyracephala rubiensis from the USA, while Kotrba (2009) described Prosphyracephala kerneggeri from Baltic Amber. A new development is that we are now in the possession of a Baltic Amber diopsid with large eye stalks for which it appears necessary to erect a new genus as a possible ancestor of Diopsinae s.s. This, in its turn would support the view to elevate Hennig's Centrioncinae and Diopsinae s.l. to family level. Furthermore, photographic evidence is available for yet another undescribed Prosphyracephala species from Baltic Amber.

Distinctive characters for Centrioncus and Teloglabrus
In Table 1, the differential character states for the genera Centrioncus and Teloglabrus are listed. The most important differences are indicated with an asterisk. Major apomorphic characters states for Teloglabrus are formed by the reduction of the anterior sclerite of female sternite 7 to two narrow elongate plates anteriorly connected on the meson, by the reduction of the posterior sclerite of female sternite 7 to two sclerites that are small to absent, and by the absence of microtrichia on the outer side of the outer and median arms of the surstylus. Major apomorphic characters states for Centrioncus are formed by the amazingly large (9.3-16.5% of body length) ejaculatory apodeme + sac and by the proximal section of ejaculatory duct turned perpendicular to the apodeme. The two genera are clearly allopatric in distribution.
McAlpine (1997) recommended treating Teloglabrus as junior synonym of Centrioncus, a view followed by Marshall et al. (2009) andLonsdale (2020). Other authors (e.g., Baker 1999;Baker et al. 2001;Meier and Baker 2002;Carr 2008;Marshall 2012;Jackson 2019;Grace and Carr 2020) used Teloglabrus without discussing its status. De Meyer (2004) discussed both genera and noted the absence of synapomorphic characters for Centrioncus. He stated that his C. bytebieri "takes a somewhat intermediate position between the two genera". Synapomorphies are now provided for Centrioncus, while C. bytebieri is confirmed as Centrioncus. Feijen and Feijen (2021), in a treatment of Afrotropical Diopsidae, provide a key to the genera and a synopsis of each genus.
Consensus on the status of Teloglabrus does not yet exist, but we think that the present paper supplies additional data to support the recognition of Teloglabrus as a valid entity. Lonsdale and Eiseman (2021) considered as main criteria for consideration in the erection of genera 1) monophyly and 2) utility. They considered that at the very minimum a genus should "represent a single monophyletic lineage that does not render other such groups non-monophyletic by its presence." Beyond the issue of monophyly, the usefulness of a higher taxon was stated to be "linked to its practical utility in diagnostics, ...". Such taxa should be ideally "diagnosed by multiple complex characters that are easily observed and present in all species". We fully agree that pragmatism should be a fundamental element in classification. For the Centrioncinae, external characters are present but subtle. However, for the two genera differences between male and female genitalia are large. The generic differences in female sternite 7 can easily be observed and do not require dissection. The giant ejaculatory apodeme + sac of Centrioncus certainly requires dissection, but is unmistakable.
In Diopsidae, external morphological differences between species or genera are often subtle. We can refer to the differences between Teleopsis Rondani and Megalabops Frey (Feijen 2011;Feijen and Feijen 2019). Besides the molecular evidence, these genera are supported as distinct by large differences in the male and female genitalia. In addition, there are substantial differences in allometric data, Teleopsis being sexually dimorphic with regard to eye span, while Megalabops is monomorphic in this regard. However, other external morphological differences certainly exist but are usually not prominent. In the rather inaccessible genus Diasemopsis Rondani, external differences are often subtle but genital differences and allometric differences are pronounced.
As far as a pragmatic system should also allow for "predictiveness" (Lonsdale and Eiseman 2021), it can be stated that although the allopatric distribution of Centrioncus and Teloglabrus does not form an argument as such for the status of these genera, it can safely be predicted that the six undescribed Southern African species that we know off, fall within the concept of Teloglabrus.

Intrageneric relationships in Centrioncus
Although the present key to species largely reflects the assumed relationships between Centrioncus species, in this discussion we will elaborate upon additional characters that were excluded from the key so as not to make it unwieldy. Based on the sets of differential characters presented in the species diagnoses, three groups of species can be distinguished. The first group can be referred to as the C. aberrans species-group. It includes C. aberrans and C. crassifemur sp. nov., and most probably C. angusticercus. Distantly related to this group are most likely C. bururiensis sp. nov. and C. decellei. It is unfortunate that in this group three species are only known from the female holotypes, and two holotypes were not available for the present study. The second group is named the C. prodiopsis species-group and comprises C. prodiopsis and C. copelandi sp. nov. The third group is the Centrioncus decoronotus group and includes C. decoronotus, C. jacobae, and C. bytebieri.
The species of the Centrioncus aberrans group (including C. bururiensis sp. nov. and C. decellei) all have a blackish brown apex on the fore femur, while the five species in the C. prodiopsis and C. decoronotus groups have a brown stripe on the inner side of the fore femur. The three core species of the C. aberrans group have a high w/l ratio of the anterior sclerite of female sternite 7 and the posterior sclerite of female sternite 7 is trapezoidal to rectangular in shape. The seven other species have a lower w/l ratio of the anterior sclerite of female sternite 7 and the posterior sclerite of female sternite 7 is distinctly U-shaped. The two species of the C. aberrans group for which the male genitalia are known, C. aberrans and C. bururiensis sp. nov., have a very characteristic shape for the surstylus (Figs 21-24, 51), with a triangular outer arm with a base much broader than the apex, and a very broad median arm. The question is whether the three species C. angusticercus, C. decellei and C. crassifemur sp. nov. also have similar male genitalia. In contrast to the males of the C. aberrans group, the males of the C. prodiopsis group and the C. decoronotus group have all a deeply constricted base of the outer arm (Figs 69, 88, 122, 138, 150) and a slender median arm. Another difference between group 1 and groups 2 and 3 can be found in the male cercus. In the males of the C. aberrans group, the cercus gradually broadens distally, but there is no abrupt distal lateral extension (Figs 17, 54). That gives for this group a cercus with a length/ greatest width ratio of 2.4-2.6. Unlike the males in the C. aberrans group, the males in groups 2 and 3 have a very distinct distal lateral extension of the cercus (Figs 71,89,123), with a length/greatest width ratio of 1.3-1.7.
Group 2, the Centrioncus prodiopsis species-group, is the most distinct one. The two species are characterised by the very large central wing spot, an absence of distal spots on the tergites, sternite 5 has a large mesal invagination anteriorly (Figs 80, 146), the female cercus has a l/w ratio of 4.0-4.4, there is an apical row of 16 or 17 tubercles on the outer arm of the surstylus, an absence of spinous setae on the median arm (Figs 88,150), and an elongate subepandrial clasper (Figs 90, 151). In contrast, the species of group 3, the C. decoronotus species-group, have a small to large central wing spot, distal pale spots on the tergites, sternite 5 has a straight anterior edge (Figs 72, 118, 132), the female cercus has a l/w ratio of 2.3-3.6, there is an apical row of 4-9 tubercles on the outer arm of the surstylus, there are 3-6 spinous setae on the median arm (Figs 69,  122, 138), and a trapezoidal to triangular subepandrial clasper (Figs 69, 124, 138).
Within group 3, Centrioncus decoronotus and C. jacobae appear to be more closely related based on the following character states: scutum with pattern of blackish brown and chestnut brown including brown humeral calli, fore femora with l/w ratio of 2.78-2.89, wing spot large and reaching crossvein dm-m in cell r4+5, sternite 4 rectangular, sternites 4 and 5 with heavily sclerotised areas, posterior sclerite of female sternite 7 without posterolateral extensions, subanal plate pentagonal, outer arm of surstylus rounded and subepandrial clasper strongly constricted basally. The same characters show the following states in C. bytebieri: scutum blackish brown including humeral calli, fore femora with l/w ratio of 3.30, wing spot small and not reaching crossvein dm-m in cell r4+5, sternite 4 trapezoidal, sternites 4 and 5 without heavily sclerotised areas, posterior sclerite of female sternite 7 with large posterolateral extensions, subanal plate triangular, outer arm of surstylus gradually tapering towards base and subepandrial clasper basally slightly constricted. These differences between C. bytebieri and the other two species appear convincing. However, there are also some similarities between C. bytebieri and C. decoronotus that are not found in C. jacobae, like larger scutellar spines (scutellar spine/length of body ratio 0.90-0.95 vs. 0.78), a shorter, narrower stripe on the fore femur, small posterolateral spots on tergite 2 (vs. large spots), and a long and slender common base of the outer and median arms of surstylus (vs. a short, broad base).

Measurements and ratios in Centrioncus
It is obvious that for differential characters in the Centrioncinae, male and female genitalia and other abdominal structures must be used. Specific differences in other morphological differences are small, although there are exceptions, such as the size and strikingly aberrant colour in Teloglabrus stuckenbergi Feijen. However, even when very characteristic scutal patterns exist, as in Centrioncus decoronotus and C. jacobae, these are often obscured by greasiness or other causes. Therefore, measurements and various ratios can give additional useful differential characters. As shown in Tables 6, 7, quite a number of these characters show distinct differences. Compared with other Diopsidae taxa, there is a remarkable lack of difference between the sexes in measurements, body ratios and number of spines and tubercles (Tables 2-5). Several of the more important differential characters regarding measurements and ratios are highlighted below.
For Centrioncus crassifemur sp. nov., its most striking external character is formed by its incrassate fore femur with a l/w ratio of 2.36. This is based on a single specimen, but is clearly distinct from the ratio in C. aberrans of 2.75 ± SE 0.01 (range 2.68-2.84, n = 22), C. bytebieri of 3.30 ± 0.02 (range 3.17-3.43, n = 16), C. decoronotus of 2.89 ± 0.01 (range 2.75-3.05, n = 43) and C. jacobae of 2.78 ± 0.02 (range 2.64-2.92, n = 18). This set of ratios also clearly shows that the fore femur in C. bytebieri is less incrassate than the other three species with large data sets. Comparing the four sets of ranges for the ratios in the four species shows that there is not even any overlap in the ratios for C. bytebieri and those of the other three species. In other taxa of Diopsidae, the l/w ratios of the fore femur have also proven to be very useful (see e.g., Feijen et al. 2018).
Examination of Tables 6, 7 clearly shows Centrioncus bytebieri as the largest species with a body length of 5.55 mm ± SE 0.09, against 5.16 mm ± 0.04 for C. aberrans, 5.15 mm ± 0.04 for C. decoronotus and 4.97 mm ± 0.06 for C. jacobae. The larger body length for C. bytebieri corresponds to other larger absolute size data such as for wing, scutellum, scutellar spine and apical seta (Tables 6, 7). However, that does not run true for all quantitative characters. The width of head is, for instance, largest in C. aberrans, while the number of tubercles on the fore femur and the number of spinous setae on the fore femur are the lowest in C. bytebieri. For relative data, useful differences can also be found. The scutellar spine, as compared to both the length of the scutellum and the body length, is, for instance, distinctly smaller in C. aberrans and C. jacobae than in C. bytebieri and C. decoronotus (Tables 6, 7).
For postabdominal characters, quantitative particulars can be very useful. The w/l ratio for the anterior sclerite of female sternite 7 forms an important differential character at species level, but also at species-group level. This ratio is high in the core species of the Centrioncus aberrans group with values between 4.4 and 8.6, while in other species the ratio is 2.5-3.9 (Table 8). In the distant relatives in the C. aberrans group, C. decellei and C. bururiensis sp. nov., the score is only 2.6-2.7. Especially C. crassifemur sp. nov. has a very wide and short anterior sclerite with a ratio of 8.6. In the C. decoronotus species-group, C. decoronotus has a high score of 3.9, while the other two species, C. bytebieri and C. jacobae have both a score of 2.6. The l/w ratio of the female cercus is another useful quantitative character (Table 8). These ratios partly correspond to the values for the ratio width/length for the anterior sclerite of female sternite 7. Centrioncus angusticercus and C. aberrans have high scores of 5.1-5.4, while the species of the C. decoronotus group have low values of 2.3-3.6. The shape of the male cercus has, in the first place, to be described qualitatively with the large and highly unusual lateral extension in the distal third in species of the C. prodiopsis group and the C. decoronotus group. Just as well, the quantitative length/greatest width ratio is also useful with values for the males in the C. aberrans group of 2.4-2.6 and values in the other two species-groups of 1.4-1.7 (Table 8).
An important quantitative character is formed by the relative size of the ejaculatory apodeme + sac as compared to the body length (Table 9). While making genital preparations, the giant ejaculatory apodeme + sac in Centrioncus can be observed with the naked eye. In Centrioncus decoronotus, the size comes close to 1 millimetre (Table  9). This quantitative character forms a main difference between the genera Centrioncus and Teloglabrus with ejaculatory apodeme + sac/length of body ratios for Centrioncus of 9.3-16.5 and for Teloglabrus of 5.6-8.4 (usually 5.6-7.3). The difference between C. decoronotus and Teloglabrus sanorum Feijen for the size of ejaculatory apodeme + sac is striking (Figs 153, 154). Feijen (1983: figs 94, 95, 115-133) illustrated the ejaculatory apodeme + sac of many Centrioncus and Teloglabrus. Now that C. aberrans has also been shown to have a large ejaculatory apodeme + sac, the large size of ejaculatory apodeme + sac can be considered an autapomorphic character of Centrioncus. Some intraspecific variation occurs for this quantitative character (see Table 9), but this can be due to the ejaculatory sac being fully extended or not. Smithers (1958) and Feijen (1983) remarked on the special tendency of Centrioncinae to discolour after death. Colour contrasts can become accentuated by preservation in alcohol. In pinned specimens the flies can become darker and "greasy" if some time lapses between collecting and pinning, especially when kept in a closed container. This tendency to become greasy has never been observed in stalk-eyed Diopsidae and might form an indication for a major physiological difference. In Centrioncus, C. decoronotus and C. jacobae can have striking pattern of brown and blackish brown on the thorax (Figs 112, 127). However, this pattern is not always visible in collection specimens, while it can even be much less pronounced in live specimens.

Colour patterns in Centrioncus
The abdomen is usually blackish brown, but can have pale posterolateral spots and/ or whitish microtrichose posterior edges. Large and striking posterolateral spots occur on tergite 2 only in Centrioncus jacobae (Fig. 135). The colour pattern of the fore femur is a useful and stable differential character. The basic colour is pale yellowish brown, but the apex can be dark brown in varying degrees or have a dark brown stripe on the inner side.
Wing patterns form a differential character in the Centrioncinae, but it should be stressed that the patterns are quite vague, certainly if compared with the often very  Feijen (1983: figs 115, 126) and are drawn to the same scale. Scale bar: 0.2 mm. dark spots in many stalk-eyed diopsids. As such, it is usually necessary to prepare wing slides to properly see the spots. In Centrioncinae, the central wing spot forms a differential character at the genus level. It is present in all Centrioncus and in only two Teloglabrus. However, De Meyer (2004) stated that no central wing spot occurs in C. bytebieri. After cursory examination of paratypes at MRAC, we initially agreed to this observation, but from examination of additional specimens it became clear that C. bytebieri has a central wing spot (Fig. 3).

Antennae in Diopsidae
For the Diopsidae, Feijen (1983: fig. 4) illustrated the antenna of Teloglabrus sanorum, Sphyracephala beccarii (Rondani) and a Diopsis of the cruciata species-group (wrongly identified as D. phlogodes). Feijen (1984: fig. 1) illustrated the antenna of a Diopsina, while Feijen (1989: figs 18, 46) provided drawings of the antennae of two Nearctic Sphyracephala Say. Feijen et al. (2018) illustrated the antenna and basal arista for six Madagascar diopsids. McAlpine (2011: figs 122-125), in an extensive study on antennal morphology in Diptera, described and illustrated the antennae of Centrioncus decoronotus and S. beccarii and briefly described the antenna of a species of Cyrtodiopsis. McAlpine described the antenna of Centrioncus as follows: "the conus [ fig. 122] is deep, somewhat bilaterally compressed, and asymmetrical, with laterally facing preapical foramen; segment 3 [ fig. 124] has its basal foramen inside the basal hollow on its lateral wall; the arista is inserted slightly laterally to the dorsal margin of segment 3." On the antenna of Centrioncinae he remarked that it is "more like that of various basal schizophoran types found in the Sciomyzoidea and Heteromyzoidea than is that of the Diopsinae, though it may partly retain the plesiomorphic structure from which that of the Diopsinae was derived." The antenna of the Centrioncinae is indeed quite different from those of the stalk-eyed Diopsidae with a pubescent arista, a distally triangular pedicellus projecting into the dorsal section of the funiculus, and the ventrally strongly extended funiculus (Fig. 38).

Facial sulcus in Diopsidae
The median suture-like groove of the facial region in many stalk-eyed Diopsidae has been referred to as the facial sulcus (Shillito 1971;Feijen 1989). According to McAlpine (1997), the facial sulcus does not divide the face on the meson, but "the facial sulcus is actually the face, which has become greatly narrowed in more advanced diopsids". The facial sulcus is absent in Centrioncinae (Figs 11, 95). In the fossil genus Prosphyracephala, no facial sulcus is found, but a large, triangular plate is present centrally in the face (e.g., Kotrba 2009: fig. 7). This plate can even be somewhat elevated. The facial sulcus is absent or reduced in several Sphyracephala and Cladodiopsis (e.g., Feijen and Feijen 2019: figs 4, 5). McAlpine (1997) assumed the "supposed absence of the sulcus in species of Sphyracephala" to be due to "smaller difference in degree of sclerotisation between parafacial and face and less narrowing of the face". McAlp-ine's explanation of the facial sulcus is very interesting but would require more study of its distribution and internal structure among the stalk-eyed Diopsidae, especially the Sphyracephalinae.
Scutal setae and supra-alar ridge (carina) in Diopsidae Feijen (1983) listed three scutal setae for Centrioncinae: a presutural seta, a supra-alar seta (SA) and an infra-alar seta (IA). However, the correct name for the presutural seta is posterior notopleural seta (PNS) according to McAlpine (1997). The IA is the largest followed by the PNS. The SA was described as rather small and standing on a small ridge. Feijen (1983) also remarked: "The SA also occurs in various Diopsina ...., although in Centrioncus this seta stands on a small ridge and as such perhaps might not be homologous to the SA in Diopsina." Feijen (1989) listed the various scutal setae in the Diopsidae. An SA is only present in three stalk-eyed diopsids: Diopsina nitida (Adams), Diopsina draconigena Feijen and Diopsina fluegeli Feijen & Feijen (Feijen and Feijen 2013). In addition, the stalk-eyed genera Teleopsis Rondani and Megalabops Frey have a pair of scutal spines which are referred to as supra-alar spines. McAlpine (1997) considered the supra-alar ridge (named carina) as a structure which is peculiar to the Syringogastridae and Diopsidae. This carina was illustrated for Centrioncus decoronotus (McAlpine 1997: fig. 35). McAlpine considered the carina "probably at its most primitive condition in Centrioncus (including Teloglabrus ...), where it consists of a well-defined ridge on the surface of the mesoscutum passing posterodorsally from the postnotopleural ridge ... to the base of the supra-alar bristle". McAlpine further stated that the supra-alar seta is vestigial or absent in Syringogastridae and most Diopsinae, but the carina remains. According to McAlpine the carina, as such, is not present in the genus Teleopsis but "the summit of the supra-alar carina is produced as a large spinous process (the supra-alar spine) and the lower part of the carina is more or less obsolete. Given the differing views on the placement of Syringogastridae and Diopsidae (Meier and Baker 2002and Marshall et al. 2009vs. Wiegmann et al. 2011, the question is whether the carinas in Syringogastridae and Diopsidae s.l. can be considered synapomorphies. Shillito (1950) in his paper transferring Centrioncus to the Diopsidae remarked "The pleurotergal spine characterizing the Diopsidae is replaced by a dome-like swelling" and in his revised family definition: "The thorax shows modification of the pleurotergite into a dome-shaped swelling or a well-developed spine". The suggestion that the swelling is likely replaced by a spine is not correct. In the stalk-eyed Diopsidae, a bulbous pleurotergite occurs just like in the Centrioncinae. In the stalk-eyed Diopsidae there is, in addition, a pleurotergal spine (sometimes called metapleural by mistake) centrally or dorsally on this bulbous pleurotergite. Size and direction of this spine are major differential characters in the stalk-eyed Diopsidae. Spine and swelling have usually been labelled as pleurotergal, but Lonsdale (2020) places these structures on the katatergite, the ventral section of the pleurotergite.

Sternopleura (= katepisternum) in Diopsoinea
McAlpine (1997) noted the pitting along the median ventral suture of the sternopleura in some taxa of Diopsoinea (Syringogastridae + Diopsidae) as an interesting character. For Syringogaster sp., McAlpine described a single median series of relatively few, spaced, deep pits. In Centrioncus spp., he found a similar condition, but the pits were more numerous. In most stalk-eyed Diopsidae the median sternopleural suture was externally visible as a groove or ridge, but without associated pits associated. Only in Sphyracephala sp., McAlpine found a series of pits along each side of a slight ridge.

Preabdominal sternites in Centrioncus
In most Centrioncus, sternite 1 is a short, rectangular sclerite with somewhat concave anterior and posterior sides. Sternite 2 is the longest sternite and is usually rectangular. In many stalk-eyed Diopsidae a tiny, strongly sclerotised sternite occurs in between sternites 1 and 2. This sclerite is referred to as intersternite 1-2. This sclerite has been regularly noted and illustrated, and was specifically discussed by Feijen (1989) and Feijen et al. (2018). In the latter paper, it was confirmed that this sclerite originates from sternite 2. Its shape can be a useful differential character even at genus-level. In most stalk-eyed flies it is not connected to sternite 2, but in Gracilopsina Feijen, Feijen and Feijen it is laterally connected to sternite 2. Feijen et al. (2018) considered the character state in Gracilopsina as the plesiomorphic state. Intersternite 1-2 is usually a small line-like sclerite with no connection to sternite 2. It can also be absent. In Sphyracephala nigrimana Loew, Feijen et al. (2018), found a situation comparable to the one in Gracilopsina, while "in two other Sphyracephala the mesal anterior edge of sternum 2 is more sclerotised". Feijen et al. (2018) stated that "In the Centrioncinae intersternite 1-2 is absent". However, this remark is now shown to be mistaken. In Centrioncus bururiensis sp. nov., C. copelandi sp. nov., C. decoronotus, C. jacobae and C. prodiopsis, spindle-shaped well-sclerotised central sections anteriorly of main sternite 2 have short lateral connections to sternite 2 (Figs 41,80,118,132,146). These sections can be referred to as intersternite 1-2. Centrioncus aberrans (Fig. 13) also has an intersternite 1-2, but lateral connections to main sternite are long. In C. bytebieri and C. crassifemur sp. nov., the well-sclerotised slender mesal anterior section of sternite 2 forms an integral part of sternite 2 (Figs 72, 101), a situation like the one in the two Sphyracephala. The state in the two Sphyracephala and the two Centrioncus can be considered the most plesiomorphic state, while the intersternites with lateral connections form the less plesiomorphic state. The presence of an intersternite, although considered apomorphic, was already verified in a Baltic amber diopsid with long eyestalks.
Sternite 3 is rectangular in most species and square in two species. Sternite 4 shows more variation and can be square, square to rounded or trapezoidal. Five species (Centrioncus bururiensis sp. nov., C. copelandi sp. nov., C. decoronotus, C. jacobae, and C. prodiopsis) have anteriorly one or two pairs of small, heavily sclerotised areas in sternite 4 (Figs 41,80,118,132,146). Sternite 5 shows also much interspecific variation. It can be rectangular, square or trapezoidal. In C. copelandi sp. nov. and C. prodiopsis, sternite 5 is anteriorly strongly invaginated on the meson, while five species (C. bururiensis sp. nov., C. copelandi sp. nov., C. decoronotus, C. jacobae, and C. prodiopsis) have anteriorly one pair of small, heavily sclerotised areas (Figs 41,80,118,132,146). Sternite 6 is in most species short, trapezoidal and broad. Centrioncus decoronotus is the only species in which sternite 6 is not so broad, while it also has an anterior pair of small, heavily sclerotised areas (Fig. 118).

Tarsal and tibial sawlines in Diopsidae
McAlpine (1997) stressed the phylogenetic importance of tarsal sawlines in the Diopsoidea. According to McAlpine, a tarsal saw line consists of a "well defined linear longitudinal series of short, compressed cuneate setulae situated on either the anterior or posterior side of a tarsal segment". For Centrioncus decoronotus, McAlpine (1997: fig. 42) presented a highly magnified (x 1740) part of the posterior sawline of the mid basitarsus. For the combination of Syringogastridae and Diopsidae, the presence of a tarsal saw line, at least on the posterior side of the mid basitarsus, was considered a synapomorphy, a view followed by Lonsdale (2020). Within the Diopsidae, McAlpine noted differences with regard to the presence of tarsal sawlines. In Centrioncus, sawlines were only present on the posterior surfaces of mid tarsal segments 1-4, extending along most of the length of the mid tibia on the posterior surface. For the stalk-eyed Diopsidae, various Sphyracephala were found to be different in the absence of sawlines on the hind tarsus, whereas other genera had at least the anterior side of hind tarsal segment 1 with a sawline. McAlpine considered the reduction of sawlines in Sphyracephala an apparently derived state within the Syringogastridae + Diopsidae. A sawline on the fore tarsus was according to McAlpine (1997) only present in, what he called, the advanced diopsine genera, like Diasemopsis Rondani, Cyrtodiopsis Frey, Teleopsis, and Diopsis.

Setal formula for legs
Feijen (1983) used a "setal formula" to describe the numbers of spinous seta and tubercles on the legs of Centrioncinae. In, for instance, Centrioncus aberrans a setal formula of 4.1, 4.9, 18.2, 17.2, 6.5 is given. The numbers refer to, respectively, mean of F1 bristles on outer row, mean of F1 bristles on inner row, mean of F1 tubercles on outer row, mean of F1 tubercles on inner row and mean of tubercles on F3. This formula represents a useful character for species. However, given the large ranges in numbers of tubercles and spinous setae (see Table 2), it must be stressed that this setal formula is only useful if based on large series of specimens. In stalk-eyed Diopsidae, the number of spinous seta and tubercles on the fore femur also forms an important character at species and genus level.

Subcostal cell
In Centrioncus aberrans, the subcostal cell is not visible in most specimens as in the flies of the type series and in all other Centrioncus. However, the subcostal cell is visible in one specimen from Rwanda and one specimen of Mt. Elgon. The visibility of the subcostal cell can therefore no longer be regarded as a reliable character as stated in Feijen (1983).

Molecular analyses for Diopsidae
Molecular systematics for Diopsidae started with the Ph.D. thesis of Baker (1999), which presented a phylogenetic analysis of 33 diopsid species using a large molecular data set. This study included representatives of Cyrtodiopsis, Diasemopsis, Diopsis, Eurydiopsis Frey, Sphyracephala, and Teleopsis, while the Centrioncinae were represented by Teloglabrus entabenensis Feijen and Teloglabrus milleri Feijen. After Baker (1999) and Baker et al. (2001), molecular studies on Diopsidae were usually based on the same data set, with additional data on some other Diopsidae species. For Centrioncinae, only Jackson (2019) added extra information for a Teloglabrus not identified to species level. To date, molecular data on Centrioncus have not been published.

Sex ratio in Centrioncus
In the Centrioncus species of which larger series are available, an even sex ratio was found: 36 ♀ and 32 ♂ in C. bytebieri, 56 ♀ and 53 ♂ in C. decoronotus and 32 ♀ and 39 ♂ in C. jacobae. For all known Centrioncus flies taken together we found 146 ♀ and 147 ♂. In the stalk-eyed Diopsidae aberrant sex ratios, usually favouring the females, are often encountered (Burkhardt and de la Motte 1985;Feijen 1989;Wilkinson et al. 1998;Paczolt et al. 2017). Burkhardt and de la Motte found that "the sex ratio of freshly emerged dimorphic flies deviated significantly from the 1:1 ratio in favour of the females" while in "cultures of the homomorphic species no significant deviations were found". Wilkinson et al. stated that "By comparing sex-ratio distributions in stalk-eyed fly (Cyrtodiopsis) progeny we found that female-biased sex ratios occur in species exhibiting eye-stalk sexual dimorphism and female preferences for long eye span." It is certainly not the case that all dimorphic (with regards to eye span) Diopsidae have aberrant sex ratios as indicated by Feijen (1989). In Centrioncus, homomorphic with regard to head morphology, no aberrant sex ratios appear to occur. Feijen (1983) discussed the presence of fungal parasites (Ascomycota, Laboulbeniales) in Centrioncus and Teloglabrus. From these two genera, only the genus Rhizomyces Thaxter was known, but the species could not be determined. Rossi and Feijen (2018)