Seven new montane species of Drosophila in the Eastern Arc mountains and Mt Kilimanjaro in Tanzania attesting to past connections between eastern and western African mountains (Diptera: Drosophilidae)

This report describes 7 new species of Drosophila found in the Eastern Arc mountains and on Mount Kilimanjaro in Tan­ zania: D. baucipyga, D. gata, D. kilimanjarica, D. neogata, D. paragata, D. pilocornuta and D. usambarensis spp.n. Two new spe­ cies complexes, the megapyga species complex (Sophophora subgenus, melanogaster group, montium subgroup) and the gata species complex (Drosophila subgenus) are introduced. Only one species, D. baucipyga of the montium subgroup, has a geo­ graphical range matching the whole Eastern Arc, from the Usambara Mts. in the north-east ofTanzania to Mt. Uzungwa in the southsouth west of the country. Five others, including one representative of the dentissima group of the Sophophora subgenus, D. usambarensis, and four representatives of the Drosophila subgenus, D. gata, D. neogata, D. paragata and D. pilocornuta, were found only in the Usambara Mts. Two of these five, D. usambarensis and D. pilocornuta were found only in West-Usambara, while two other related species, D. gata and D. paragata, were found only in East-Usambara. Only the distribution of D. neogata covers the whole of the Usambara mountains. Outside the Eastern Arc, another representative of the dentissima group, D. kilimanjarica, was found only on Mount Kilimanjaro. This new, highly specific, montane fauna of Drosophila further contributes to the unique bio­ logical diversity of the Eastern Arc Mts. The biogeographic affinities of the new taxa suggest past connections with the Virunga and Ruwenzori ranges and further west with the Cameroon Volcanic Line. It indicates, in particular, that the Eastern Arc forests have passed through a succession of coalescence and fragmentation events.


INTRODUCTION
The Eastern Arc mountain forests of Tanzania in East Africa is a range of fragmented residual submontane rain forests extending from the Indian Ocean coast at the north-easternmost border with Kenya to the northern shore of Lake Malawi.Miocene uplift of the central African plateau separated these coastal and montane for ests from the main Guineo-Congolian forest of west and central Africa.Since then, the remarkable stability of the Indian Ocean currents that brought moisture to the tropical East African coast maintained, in the region, high rainfall throughout the Pleistocene droughts that caused the shrinkage of the African rainforest elsewhere on the continent.As a result, patches of the former Pan-African rainforest have survived on the east African mountains (Lovett, 1993a;Griffiths, 1993;Lovett & Wasser, 1993).Unlike the vast West and Central African forests, these highly fragmented forest relics are similar to intraconti nental islands associated with localised areas of high rain fall, contrasting with the surrounding arid woodland.This NE-SW mainland "archipelago" includes ancient, Terti ary, crystalline mountains, most notably Mounts Usam bara, Mt Guru, Mt Uluguru and Mt Uzungwa.Unlike the Eastern Arc mountains, Mount Kilimanjaro is a far more recent, Pleistocene, volcanic mountain, probably less than 1 My old (Wilkinson et al., 1986;Griffiths, 1993).
The antiquity, isolation and fragmented nature of the eastern African forests have resulted in remarkably high levels of endemism and diversity (Lovett & Wasser, 1993).Although great emphasis has been placed on the exceptional fauna and flora, including arthropods of the Eastern Arc and Mt Kilimanjaro (Kielland, 1990;De Jong & Congdon, 1993;Hoffman, 1993;Scharff, 1993), little attention has, as yet, been paid to the fauna of Drosophili dae.Since Lovett and Wasser's (1993) volume devoted to the biogeography and ecology of the rainforests of eastern Africa, one of us (D.L.) has made two Drosophila field surveys in Tanzania in 1995 and 1996.They revealed that the drosophilid fauna is as rich and unique as the other faunas and floras previously reported from the region, probably because of the fragmented nature of the rainfor ests and their varying antiquity.This paper describes seven new montane species o f Drosophila from Tanzania, six from the Eastern Arc forests and one from the Mt Kilimanjaro forest.Their biogeographic affinities indicate past "montane" forest corridors linking the east African mountains with the western rift, Cameroon and Guinean highlands.
We have used the spelling Uzungwa and Kilimanjaro instead of Udzungwa and Kilimandjaro throughout this paper.
Definition.The new Drosophila megapyga species complex is characterized by the following traits: two long sexual combs, one on basal (20-34 teeth) and the other on second (14-20 teeth) fore tarsomeres; cercus bearing 4-5 huge strong teeth protruding from the male terminalia, about 1/20 the body size of the male; hypandrium with a long, marked medial extention sides of which are hirsute basally as well as the inner side of the lateral extensions of the novasternum; two paramedian short, strong bristles set apically on the medial extension; anterior parameres well differentiated and easily visible through the marked depressions between the medial and lateral extensions of the novasternum; posterior parameres almost as tall as the distiphallus.The complex includes three species: D. baucipyga Lachaise & Chassagnard, sp. n., this work;D. eupyga Tsacas, 1981;D. megapyga Tsacas, 1981 Head.Frons brownish-yellow, narrow yellow band ventrally; frons width and height subequal, fw/fh -1.1-1.2;frons width/head width (fw/hw) -0.43-0.45;orbital plate same colour as the frons but glossy; or1 slightly longer than or3, or2 ca.V length of or3 and arising closer to or1; proclinate orbital/posterior reclinate orbital (or1/or3) -1-1.27,anteriorreclinate orbital/poste rior reclinate orbital (or2/or3) -0.4-0.54;ocellar triangle reddish.Face: glossy greyish-white; facial carina narrow, ca.2/3 length of face, with a white stripe on the edge; arista with 4-5 dorsal and 3 ventral rays plus terminal fork; 2nd peristomal bristle as long as the vibrissa; clypeus same colour as frons; palpus with one apical seta; gena narrow and yellow; genal width in line of greatest diameter of eye/greatest diameter of eye (g/e) -0.04.
Abdomen.Glossy; tergites 2 to 5 yellow with a narrow band posteriorly, brown for the first four tergites, black for the fifth, tergite 6 entirely black.
Distribution.Tanzania (Eastern Arc Mountains).epandrium over nearly half its length), which does not fit any of the species complexes described by Tsacas (1980).
Male terminalia.Epandrium and cercus brownish black, glossy; lateral expansion of epandrium appearing narrow and finger-like in caudal view, wide dorsally and markedly narrowed in lateral view, a row of setae along the posterior edge of the epandrium; cercus short, triangu lar, markedly protruding posteriorly, and fused to epandrium in dorsal position, lower surface of cercus densely pilose; Surstylus finger-like in caudal view, rectangular in lateral view, bearing a strong black tooth subapically and three short apical setae (Figs 7, 8); Hypandrium with a short novasternum, considerably overtopped by distiphallus and anterior parameres, without medial indenta tion in between the paramedian setae; anterior parameres huge, sabre-like and dark; posterior parameres united but not fused, hidden behind aedeagus in ventral view, also sabre-like but half as long as the anterior ones; Aedeagus long, straight; slightly lozenge-shaped in ventral view, distiphallus swelling apically in lateral view; aedeagal apodeme short and wide (Figs 9, 10).Etymology.Allusion to the origin of the specimens from Mts Usambara.
Male terminalia.Epandrium and cercus brownish black, glossy; cap-shaped in lateral view that is very large dorsally and tapering ventrally; a fringe of long setae along the posterior edge and the lateral expansion of the epandrium; cercus not fused to epandrium and most remarkable in being made up of two lobes separated by a deep medial indentation, the ventral lobe racket-like and bearing ca. 25 short strong teeth evenly distributed; surstylus elongated bearing a dorsal row of 3 prensisetae pointing downwards, a ventral row of 5 prensisetae pointing inwards, and a ventral inner lobe with 7 apical setae (Figs 13,14).Hypandrium with a rectangular novasternum the height of which exceeds the width, with almost no medial indentation between the paramedian setae; anterior parameres exceptionally stout and dark, in ventral view suggesting the dorsal part of a crab claw with outer serration dorsally and inner serration medially; posterior paramere entirely fused into a high collar hirsute ventrally, very large in lateral view.Distiphallus prickly and hook-like in lateral view; aedeagal apodeme long, narrow and incurvated (Figs 15,16).Etymology.Allusion to the origin of the specimens from Mt Kilimanjaro.
subg.Drosophila Fallén, 1823 virilis section gata complex, n.Definition.The Drosophila gata species complex belongs to Drosophila s. str.and can be arranged tenta tively into the virilis section of the subgenus.Within the section, it shows general similarity with the Palaearctic D. picta Zetterstedt, 1847, as redescribed by Tsacas (1969), including more especially pleura yellowish with three brown stripes and symmetrical black spots on tergites.However, there is a more or less distinct medial line of abdominal spots in D.picta.
Epandrium elongated and hirsute; cercus hirsute and partially fused to epandrium at a mid dorso-ventral posi tion of both cercus and epandrium; surstylus with 7-10 long and strong prensisetae; hypandrium short and rounded; distiphallus cornet-like evocative of a cat head in ventral view; novasternum hirsute posteriorly; with a marked paramedial indentation delineating two pairs of lobes, the inner one invaginated in the larger one and bearing one strong medial bristle; distiphallus large, with serrated "ears".Includes Differential diagnosis.Species with yellowish pleura crossed by three faint diffuse brown longitudinal stripes; abdominal tergites yellowish with a brown patterning gradually fragmenting antero-posteriorly; cercus partially fused to epandrium by a narrow lateral bridge at a mid dorso-ventral position of both cercus and epandrium; Male terminalia characterized by surstylus bearing a mar ginal row of 7-8 large, pointed prensisetae; aedeagus short and stocky; distiphallus cornet-like evocative of a cat head in ventral view, serrated for the most part.
Thorax.Scutum and scutellum yellowish-brown, darker in between dorso-central setae and scutellar setae; basal and apical scutellar setae subequal, b/a = 0.91, the former convergent, the latter faintly convergent; anterior dorsocentral setae shorter than posterior ones; acrostichal setulae in about six rows.Pleura yellowish with three faint diffuse brown longitudinal stripes, the uppermost along the notopleural suture, the mid one across anepis ternum and anepimeron, extending at the border between katatergite and anatergite and further on the base of the halter, and the lowermost stripe short, making an angle and fringing the upper katepisternum anteriorly; two major katepisternal setae, the anterior about 2/3 the poste rior seta, a/p = 0.66, the third mid one reduced to a setula.Legs pale.Wing length = 2.05 mm; ww/wl = 0.4; C = 2.15; C3 fringe = 40.
Abdomen yellowish with symmetrical brown patterns separated by a yellow median band: T1 without pattern, T2 with crescent-shaped spots, T3-T5 with triangular pat terns, T6 with a pair of spots, T3-T5 with additional Etymology.From modern Greek yaxa = cat, allusion to the cat-like shape of distiphallus in ventral view.
Distribution.Tanzania (East-Usambara).-medial indentation of hypandrium more markedly wid ened anteriorly, and the two lobes less rounded; -distiphallus with two strongly serrated and denticu lated dorsolateral expansions, showing in lateral view a small notch between the serrated and unserrated apical parts and with a patch of excrescences dorsally.
Abdomen.Yellowish with a dark brown longitudinal dorsal band gradually breaking up along the antero posterior axis into two spots on T4 and four spots on T5-6.T2 is also partially interrupted anteriorly; T2-6 with a dark-brown fringe laterally.
Male terminalia.Epandrium oblong, bearing 8-9 long setae along the posterior edge of the lateroventral expan sion.Cercus fused to epandrium, with long cercal setae becoming shorter on the ventral edge; cercus and epandrium densely but unevenly pubescent.Surstylus bearing a characteristic wide array of 9-10 prensisetae and a strong protruding ventral lobe bearing 3 short and 3-5 longer setae apically (Figs 21,22).Hypandrium short, with a medial indentation more markedly widened anteri orly than in D. gata, the two lobes less rounded than in D. gata.Anterior parameres pubescent, coupled to hypan drium inner edge of which is also pubescent.Aedeagus short; distiphallus with two strongly serrated and denticu lated dorso-lateral expansions, showing in lateral view a small notch between the serrated and unserrated apical parts and with a patch of excrescences dorsally.Aedeagal apodeme swollen anteriorly in lateral view (Figs 23,24).

Drosophila paragata
Thorax.Scutum and scutellum yellowish-brown with diffuse darker pattern; Anterior dorsocentral setae shorter than posterior ones; six rows of acrostichal setae; basal and apical scutellar setae convergent.Pleura yellowish, with three brown stripes very similar as in D. gata; two major katepisternal setae, the posterior stronger than the anterior one, a/p = 0.52, and one additional minute setula in mid position.Legs pale.Wings greyish and iridescent, wing length = 1.97, ww/wl = 0.40; C = 2.10; C3 = 36.8.Abdomen.Yellowish with a dark brown longitudinal dorsal band interrupted in mid-dorsal line resulting in two symmetrical brownish patterns very similar to D. gata but paler.
Male terminalia.Epandrium oblong bearing 6 long setae along the posterior edge of the lateroventral expan sion and a few shorter setae apically on this expansion.Cercus fused to epandrium, with long cercal setae becoming shorter on the ventral truncated edge.Cercus and upper part of epandrium densely pubescent.Surstylus bearing a slightly convex row of six prensisetae and a strong protruding lobe bearing 2-3 short and 3-4 longer setae apically (Figs 29,30).Hypandrium rounded, with a notch-shaped medial indentation; the posterior margin of hypandrium protruding into two more medial lobes, pubescent, bearing one paramedian seta.Aedeagus short; distiphallus with two long apical expansions faintly ser rated and separated from one another by a bare gently undulating border, a patch of excrescences can be seen dorsally in lateral view (Figs 31,32).Etymology.From the modern Greek reapa = "close to", evoking the relatedness with D. gata.
Distribution.Tanzania (East-Usambara).Although more or less dis tantly related to species of the gata species complex, it differs by its ringed femora and the lampbrush-like pro jections of the distiphallus.
Thorax.Scutum and scutellum yellowish-brown, three faint, diffuse darker longitudinal bands visible under some views: two lateral bands ending at the level of ante rior dorsocentral setae, and one medial band reaching the scutellum where it divides into two larger bands.Anterior dorsocentral setae shorter than posterior ones; acrostichal setulae in six rows; basal scutellar setae almost parallel, apical ones stronger and convergent, b/a = 0.87.Pleura yellow with two faint longitudinal brownish stripes: the uppermost running step-like across upper third of anepis ternum, and then across mid anepimeron, lower third of katatergite + anatergite and upper border of katepisternum; two major katepisternal setae, the anterior shorter than the posterior one (a/p = 0.52), plus a minute setula in a medial position.Legs pale yellow, mid femora with a faint apical brownish ring, hind femora with two (subbasal and subapical) brownish rings, hind tibia with a small basal brownish ring.Wings greyish and iridescent, wing length = 1.58 mm, ww/wl = 0.43; C = 2.5; C3 = 35.7.
Abdomen.Yellow with a dark brown longitudinal dorsal band interrupted in mid-dorsal line resulting in two transverse extensive spots on T2, four large triangular spots on T3, two large saddle-shaped marks on T4, and two smaller spots on T5 and T6; T1 to T4 with two small marginal brownish spots.
Male terminalia.Epandrium as wide as high with a lateral expansion square in lateral aspect bearing 5-6 long setae on the ventral edge and 1-2 others more dorsally.Cercus widely fused to epandrium, densely pubescent as well as epandrium dorsally, and uniformely covered with setae of moderate size.Surstylus bearing a row of 6-7 strong blunt prensisetae accompanied by 4-5 apical and 2-4 inner setae (Figs 33,34).Hypandrium bare showing a typical " U ''-shaped ventral pattern, very narrow anteri orly with a posterior edge markedly denticulate, the most prominent denticles bearing a long paramedian setae; anterior parameres cryptic, rounded in ventral view and elongate in lateral view, with 2-3 sensillae.Aedeagus long; medially swollen; distiphallus hairy suggesting a double flexible lampbrush; aedeagal apodeme narrow and bent (Figs 35,36).Etymology.From Latin, evoking the lampbrush of the disti phallus.

DISCUSSION
The Eastern Arc forests harbour a range of endemics in different forest types and of different affinities and there fore it is thought that the Eastern Arc cannot have suf fered a major loss of forest or a sustantial lowering in the altitude of forest types, at least in the last 30 000 years (Lovett, 1993b).The centres of endemism probably resulted from the persistence of moist forests during the Pleistocene droughts.In particular, the Usambara moun tains contain the highest known diversity and endemism of plant and animal species in East Africa (Newmark, 1991).The seven remarkable new species of Drosophila found in the Eastern Arc mountains and on Mount Kilimanjaro and described in the present work provide futher evidence that the highly fragmented montane for ests ofTanzania have long been centres of speciation.
The new species, D. baucipyga, is undoubtedly among the most spectacular representatives in the Afrotropical region of the basically Oriental montium species sub group.It is also most interesting for its biogeography.The geographical range of D. baucipyga is at once local ized and widespread, matching the range of the suppos edly old forests (in a geological sense) of the overall Eastern Arc, from Mount Usambara to the north-east of Tanzania to Mount Uzungwa to the south-south west of the country.Otherwise, the species is seemingly absent from the much younger forests of Mt Kilimanjaro and Mt Meru (D.L. unpublished).Should this distribution be con firmed, it probably means that the species has little pro pensity to migrate and the subdivided extant populations result from the fragmentation of the species home range.If this is so, D. baucipyga would be a palaeoendemic, a relic of the formerly continuous submontane Eastern Arc forests.This would support further the view that these ancient crystalline mountain forests have been intercon nected, and isolated from distant forest blocks for a rela tively long period of evolutionary time (Lovett & Wasser, 1993).Worth noting in this respect is the striking simi larity of D. baucipyga to the species pair D. megapyga from Ruwenzori and the Cameroon Volcanic Line, and D. eupyga from the Cameroon highlands and Gabon (Tsacas, 1981).The three species, D. baucipyga, D. megapyga and D. eupyga of the montium species subgroup all have very similar overall phenotypes, sexual combs on the basal and second fore tarsomeres, and similar -albeit distinctepandrium and hypandrium.Noticeably, all three species have a row of 4-5 huge cercal teeth protruding from the male abdomen, making them very easy to identify.They are undoubtedly closely related and we propose to place them all in a new species complex, the megapyga species complex.The finding of D. baucipyga in the Eastern Arc forests is of great interest for understanding biogeog raphical patterns throughout tropical African mountains, since it clearly indicates an old connection between Usambara, Ruwenzori and the Cameroon highlands.The presence of D. eupyga at lower elevations in northeastern Gabon (500 m a.s.l. at Makokou), and at mid (800 m a.s.l. at N 'Kolbisson by Yaounde) and higher elevations (2000 m at Bafut N 'Guemba on Mt Lefo) in Cameroon indicates that these basically montane species could migrate between forest blocks.
Two of the new Drosophila species described here belong to the dentissima species group, as revised by Tsacas (1980).This group of strictly tropical African spe cies is basically comprised of "montane" species, whose range lies mostly between 1400 and 2750 m (a.s.l.).Although the core distribution is throughout the East African mountains (Virunga, Ruwenzori ranges, and Kenyan highlands), there are notable radiations in the Cameroon Volcanic Line and a few outlying species can be found still further west on the remote Mt Nimba in Guinea.To the south, species of the dentissima group have been recorded on the Nyika and Viphya plateaux in northern Malawi and further south in the eastern escarp ment of Zimbabwe and those of the Drakensberg in South Africa (Chassagnard et al., 1997).A few rare species extend to lower elevations up to the Cape peninsula in South Africa.
The Drosophila dentissima group is probably one of the major groups of invertebrates shedding light on the palaeogeography of the East African highlands and their connections with the Cameroon highlands.Apart from a few rare outlying species, the core distribution of the den tissima species group matches what has long been called the "Victorian areotype" (Sharpe, 1893) (by reference to Victoria city by Mt Cameroon, Lake Victoria in Kenya/Tanzania, and Victoria Falls in Zimbabwe).The two new species described here are the first records of dentissima relatives reported from the Eastern Arc moun tains and from Mt Kilimanjaro, and more generally from Tanzania.
Drosophila usambarensis, found in two locations of West-Usambara (Mazumbai and Mt Magambo), contrib utes to the uniqueness of Mt Usambara.The species cannot be ascribed to any known species complexes of the dentissima group recognized by Tsacas (1980).It exhibits a combination of traits so far unknown and par ticularly extraordinary, including a long sexual comb of basal tarsomere; one long subapical oral seta; united but not fused posterior parameres; and the cercus fused to the epandrium over nearly half its length.This last major character has not been seen in the dentissima group.It is probably the first representative of a new species complex.However, we will delay the description of this latter until related taxa are found.
Outside the Eastern Arc, another representative of the dentissima group, D. kilimanjarica, is only reported from the relatively recent forests of Mount Kilimanjaro.The species has a combination of quite original traits, including a long sexual comb of basal tarsomere; two long subapical oral setae; posterior parameres entirely fused; and a cercus not fused to the epandrium.The new species could be tentatively assigned to the altissima spe cies complex, differing by a more complete fusion of pos terior parameres from D. kilimanjarica.The altissima species complex is represented by D. altissima Tsacas, 1980 from the Aberdare range in Kenya and the Virunga range in Zaïre, and D. oreia Tsacas 1980 from the Ruwenzori range in Uganda (Tsacas, 1980).The altitu dinal range of the altissima complex is 1850 to 2750 m, possibly more, making this cluster of taxa a strictly mon tane species complex, the highest known in the tropical African region.
The Drosophila subgenus (i.e., Drosophila s. str.) pro vides four other new taxa with restricted geographical ranges within the Eastern Arc mountains, in addition to D. usambarensis.These are D. gata, D. neogata, D.paragata and D. pilocornuta, which have only been reported from Mount Usambara.Only one of these four species, D. pilocornuta, is known to date, like D. usambarensis, from West-Usambara, while two other related species, D. gata and D. paragata, were only recorded from East-Usambara.Uniquely, D. neogata, has a distribution that covers all the Usambara mountains.Three of these four new species, D. gata, D. neogata and D.paragata, appear to be closely related on the basis of overall and detailed morphology and we propose to place them in a new cluster of taxa, the gata species complex.The assignment of this new species complex to a species group is more questionable (the gata species complex is nonetheless tentatively arranged here into the virilis section of the subgenus), a difficulty that was stressed formerly for other new complexes (Tsacas & Chassagnard, 1999) & Chassagnard, 1994& Chassagnard, , 1999& Chassagnard, , 2000, and present data), and present data).This new fauna of Drosophila contributes further to the great uniqueness of the Tanzanian mountains and sug gests a staggering of speciation events depending on the more or less advanced fragmentation and isolation of the covering submontane forests.These new findings provide further evidence that the Usambaras shelter the greatest number of species probably because of their favoured position as a coastal rain-trap (Kingdon, 1990).This author argued that such diversity is primarily due to their being the moistest and most favourable region along the whole eastern seacoast.But the unique richness and diversity of the Usambaras is also reinforced by isolation, the nearest blocks of rain-trapping mountains lying not only 130 km away to the north and south but further inland.
Ecological data may provide further indications about the distribution and diffusion of "montane" Drosophila species.The Mazumbai forest can be seen as 'an island of lower montane rainforest in the West Usambara' (Red head, 1981).There are four major types of forest at Mazumbai: Ocotea/Podocarpus; OcotealSyzygium', NewtonialSyzygium', and, Newtonia/Parinari/Syzygium (Hall, 1985).The Newtonia/Parinari/Syzygium forest appears to be most suitable for Drosophila.While the pods of the legume Newtonia (Mimosoidea) are unsuit able for Drosophila, the fruits of both Parinari excelsa Sabine (Chrysobalanaceae) and Syzygium guineense (Wild.)DC. (Myrtaceae) yield a great diversity of drosophilid species, including the two new species described here, D. baucipyga and D. usambarensis.According to Lovett (1993b), Parinari excelsa is the most common species of tree in the submontane forest in the West Usambara: it accounts for the greatest number of indi viduals -(c.7%) and is the most frequent species, occur ring in the greatest number of plots (55%).It is also the species that has the largest basal area (c.13%).In con trast Syzygium guineense is not considered to be among the ten most common and frequent trees in West Usam-bara in general, although locally abundant at Mazumbai (Lovett, 1993b).It could still be however among the most preferred host plants for Drosophila (Table 2).The find ings of Lovett (1993b) indicating that Parinari excelsa is more common (c.20%), frequent (c.90%) and contrib utes the greatest basal area (c.47%) in northern Uzungwa in the Mwanihana Forest Reserve is consistent with this expectation.There, Syzygium guineense is also among the ten most common (3%), frequent (34%) and contributive species (2%).D. baucipyga was recorded on P. excelsa fallen fruits in West-Usambara and also occurs in the Mwanihana Forest Reserve.It is therefore likely that this species is closely associated with these host-plants.
In addition to these species, fig trees are major breeding sites for Drosophila in Tanzania in general and more especially in the Usambaras.Of a total 40 species of Ficus reported from Tanzania (Berg, 1990;Berg & Wiebes, 1992), 19 were recorded during our 1995-1996 investigations in submontane forest habitats.Sixteen of these species provided figs, of which nine yielded Droso phila.Both Ficus mucuso Ficalho and Ficus sur Forsskâl produce huge numbers of syconia (figs) which mature, and hence fall, synchronously.Such resource patches on the soil are very attractive to a diversity of unspecialised Drosophila differing ecologically from specialized, strictly fig-breeding drosophilids (Harry et al., 1996(Harry et al., , 1998)).As a result, three new species were found sympatrically on fallen F. mucuso syconia, namely D. baucipyga, D. neogata and D. gata in Amani in East-Usambara.In West-Usambara, D. neogata was caught on fallen syconia of Ficus sur and D. pilocornuta on fallen syconia of Ficus chirindensis C.C. Berg, a fig tree restricted to East Africa (Table 2).
The megapyga species complex of the montium sub group and the dentissima group all attest to the existence of past connections between the forests of the Eastern Arc mountains, the Virunga and Ruwenzori ranges and those of the main western Guinea-Congo forests, particularly those of the Cameroon Volcanic Line.For the dentissima group alone, connections extend still further west to the Guinean range (Mt Nimba).The persistent riverine for ests of the Congo river may have been a link between the Virunga-Ruwenzori forests and north-eastern Gabonsouthern Cameroon region.The genera Parinari, Syzy gium and Ficus are widespread along this putative north-Congo migratory pathway and may have been among the host-plants that repeatedly favoured the migration of Dro sophila during various geological and climatic episodes.

Table 1 .
. If Position of the seven new Drosophila species in the within-genus classification and their distributions in Tanzanian mon tane forests.Asterisks indicate new taxa; (-) = no subgroups recognized in the relevant species group; (likely new) = the description of which will be delayed until related taxa shall be found.

Table 2 .
Resources on which new Drosophila species were recorded in Mt Usambaras, including the fruits of the forest trees of the genera Parinari (Chrysobalanaceae), Syzygium (Myrtaceae) and Ficus (Moraceae).No resources are reported for D.paragata.