Sapho orichalcea – This species has been separated from S. gloriosa McLachlan, 1873 by the absence of a matt band on the otherwise wholly dark wings of both sexes (Pinhey 1971b). However, this band’s extent and position is variable and the validity of these species in untested (Dijkstra unpub.). Angolan populations lack the band, consistent with S. orichalcea as currently defined.

Chlorocypha and Platycypha – Platycypha Fraser, 1949 is distinguished from Chlorocypha Fraser, 1928 only by the largely coloured and often widened male tibiae (Dijkstra and Clausnitzer 2014). However, COI sequences indicate that the blue Angolan species classified as Chlorocypha (bamptoni, crocea, rubriventris and a possible fourth species) and Platycypha angolensis (formerly a subspecies of P. caligata, see below) are closer to each other than to any Chlorocypha species, grouping nearer Platycypha (Dijkstra unpubl.). While P. angolensis (Fig. 6a) has legs typical of Platycypha, the blue Angola “Chlorocypha” males have black tibiae with white anterior faces, found also in many Chlorocypha species, but lack the yellow to red tibial markings found in other Platycypha. However, this area is paler than the femora and tarsi in females and fresh males of bamptoni and crocea. While the tibiae appear slender, those of the three species examined are actually expanded slightly and more than in Platycypha picta (Pinhey, 1962) and P. rufitibia. The black on the dorsum of the second abdominal segment (S2) shifted centrally onto the segment and covering its full extent also recalls typical Platycypha. Females have the black line on the entire dorsal abdominal carina present in all Platycypha but rare in Chlorocypha. The male of rubriventris has a red wash on the sides and underside of the thorax and abdomen, which is common in Platycypha but not found in Chlorocypha. Finally the resting behaviour with exposed tibiae as shown in Figure 5b is also more typical of Platycypha than Chlorocypha (A. Günther in litt.). It thus seems warranted to transfer P. bamptoni, P. crocea and P. rubriventris to Platycypha (new combinations). Separating Platycypha may make Chlorocypha paraphyletic, but the species transfer itself does not affect that.

Platycypha angolensis and P. caligata – While described as a subspecies of P. caligata, COI sequences of angolensis are nearer (but not identical) to those of crocea and cf. bamptoni than typical caligata from eastern and southern Afrca. Moreover, angolensis males have the thorax sides greenish with hardly a red tinge at maturity, the fore tibiae more abruptly widened at the base, the (especially hind) femora apically reddish and not contrasting with the tibiae, and black apical bars present on the abdominal segments up to S6 rather than S4 or S5. P. angolensis is thus treated as good species, which is widespread on and around the Angolan plateau. True P. caligata appears to extend into Angola from Zambia.

Platycypha bamptoni, Platycypha cf. bamptoni and P. crocea – The former was described as a subspecies of the latter but is easily separated in the field by its larger size, more robust build and distinct thoracic and abdominal markings. We propose to treat them as distinct species. Specimens like P. bamptoni from the upper Cubango differ from those of Tundavala and might represent an undescribed species.

Platycypha rubriventris – This species is only known from two males collected in 1965 at Teixeira de Sousa in north-eastern Angola. This should refer to the modern town of Luau on the DRC border, although its altitude (about 1,000 m) and location well to the east of the central plateau are surprising for an Angolan endemic.

Elattoneura cellularis, E. frenulata and E. tarbotonorum – Pinhey (1961b) reported a specimen from Lóvua near the Zambian border in eastern Angola as E. frenulata. The latter is confined to the Cape of South Africa, an error he corrected with the description of E. tropicalis Pinhey, 1974. That species, common in Zambia adjacent to Angola, was synonymised with E. cellularis by Dijkstra (2007b). Pinhey’s (1975) record of E. frenulata from Serra da Chela, however, pertains to the recently described Angolan endemic E. tarbotonorum.

Elattoneura cf. glauca – Dijkstra et al. (2015) showed that E. glauca is genetically heterogeneous, specimens from Angola being most distinct and closer to specimens identified as E. cellularis from Katanga and Zambia. E. glauca males collected near Kinshasa, directly north of Angola, differ in penis morphology from typical East African males (Dijkstra unpubl.). As with other widespread species of southern and eastern Africa like Platycypha angolensis and Nesciothemis farinosa (see below), the geographically relatively isolated populations in the west may represent a distinct taxon.

Mesocnemis cf. singularis – Longfield (1947) and Pinhey (1965) reported a female of M. singularis from near Cubango in 1932. However, we doubt the identification because we found a species at localities across Angola that differs from typical M. singularis by the COI sequences and male appendages. Similar males are known from Gabon and Cameroon. Preliminary data suggest these and presumed M. singularis form a complex of three or more species (Dijkstra unpubl.).

Agriocnemis cf. maclachlani – Females in this genus are difficult to identify morphologically. The COI sequence of one collected near Uíge is nearest (but not identical to) that of males of A. maclachlani from central and western Africa.

Crenigomphus cf. cornutus – Females of this genus are difficult to identify. Those collected recently near N’dalantando and Cuchi are morphologically similar to females caught in copula with males of C. cornutus in Zambia close to the Angolan border.

Mastigogomphus chapini – Pinhey (1961b) described Microgomphus bivittatus from a female from Dundo. Dijkstra (2007a) suggested that it might belong to Lestinogomphus, but Pinhey’s drawings actually match the female of Mastigogomphus chapini and is considered synonymous (Dijkstra et al. 2015).

Microgomphus cf. nyassicus – Microgomphus identification is uncertain as long as the status of the potentially synonymous M. schoutedeni Fraser, 1949 is unresolved.

Notogomphus cf. spinosus – The larvae of this genus have not been treated taxonomically. The COI sequences of two collected near Uíge are nearest (but not identical to) those of six adults of N. spinosus from Gabon.

Onychogomphus cf. styx – The taxonomy of the Afrotropical species placed in Onychogomphus is problematic, especially the dark species (Dijkstra 2007a). O. seydeli (Schouteden, 1934) from Zambia and Katanga and O. styx known mainly from Uganda cannot be separated reliably at present (Dijkstra and Clausnitzer 2014). Identification of Angolan specimens thus awaits revision of the group.

Paragomphus cognatus – Despite progress made by Dijkstra et al. (2015), the taxonomy of the cognatus-group of Paragomphus is problematic: especially older records of P. cognatus cannot be considered reliable, such as a female from Ebanga (Longfield 1945). Longfield (1959) stated that a male P. cognatus from Dundo differed from (topotypical) South African specimens by its markings, suggesting it belonged to another species. Pinhey (1961a) described P. machadoi from the same collection, but the holotype’s collection dates differ from Longfield’s (1959) male. Pinhey also mentioned P. alluaudi (Martin, 1915) from Dundo, but that is an eastern African species. Possibly he had the similar P. abnormis (Karsch, 1890) before him, which we recorded recently near Uíge. True P. cognatus is known from the Okavango and Cunene Rivers in northern Namibia and thus expected in southern Angola (Suhling and Martens 2007).

Paragomphus cf. darwalli and P. cf. nigroviridis – The larvae of this genus have not been treated taxonomically. The COI sequence of one larva collected near Uíge is nearest (but not identical to) that of ten adults described recently as P. darwalli from Gabon, while another from there is nearest (but not identical to) eight adults identified as P. nigroviridis from Liberia, Gabon and DRC (Dijkstra et al. 2015).

Aethiothemis mediofasciata – This species was described from a single female collected by Monard at Chimporo in southern Angola in 1928. While the holotype agrees with females associated with a species found from Liberia to Gabon and Congo-Brazzaville, the recent rediscovery of A. gamblesi (Lieftinck, 1969) in Zambia casts doubt on its taxonomic status. Males are almost identical, the abdomen being orange with maturity in A. mediofasciata (as now known) and grey pruinose in A. gamblesi. No characters to separate females are known as yet. The Angolan type locality is highly isolated: 1,300 km south of the nearest recorded orange male and 900 km west of the nearest pruinose male. Likely only further records of males from Angola can resolve the issue.

Nesciothemis cf. farinosa – Large black-tipped Nesciothemis are currently classified as two species that differ mainly in the extent of pruinosity on the mature male abdomen: at most to the base of S4 in N. pujoli Pinhey, 1971 from western Africa, but up to the very base of S7 in N. farinosa from southern and eastern Africa, although usually it does not extend beyond S5 and (more rarely) S6. However, in Angola and adjacent Namibia and Botswana mature males invariably have even the entire dorsum of S7 pruinose and sometimes the extreme base of S8 as well. COI sequences of Angolan specimens differ more from those of both N. pujoli and N. farinosa sampled across Africa, than these two differ from each other (Dijkstra unpubl.). Thus, an unnamed taxon may be present in south-western Africa.

Trithemis cf. dubia – Males from eastern Angola and Zambia differ in the shape of the hamule from those collected in Gabon, while the thoracic pattern differs from the female holotype of T. dubia from Lake Asebbe in Gabon. They belong to the longistyla-group of Trithemis with several species placed formerly in Anectothemis, Congothemis, Lokithemis and Porpacithemis (see Dijkstra et al. 2015).

Trithemis imitata and T. monardi – It is unclear whether these are distinct species or if T. imitata is a paler and generally more northerly variety of T. monardi. Both are currently listed for Angola.

Trithemis palustris – Damm and Hadrys (2009) separated two species from T. stictica (Burmeister, 1839) on genetic and morphological grounds. However, they only provided genetic characters to distinguish T. palustris from its alleged sister species T. morrisoni Damm & Hadrys, 2009. Our records may thus represent both taxa, but given the limited geographic scope of the 2009 study, these may also be synonymous.

Angola’s rich dragonfly fauna expresses its geographic position, size and diversity (Fig. 2). Dijkstra et al. (2011) observed that roughly half of tropical African species occur predominantly within the extensive lowland forests of the western and central continent, a quarter is associated with the eastern and southern part dominated by highlands, while the remaining quarter occurs in open habitats throughout much of the Afrotropics. Indeed, about half of Angola’s species are widespread across the continent and its exceptional diversity can be attributed to two major sources:

Almost 30% are confined to forest habitats in the north, mostly below 1,000 m asl, e.g. Phaon camerunensis, Umma longistigma, Chlorocypha cancellata, Allocnemis pauli, Elattoneura lliba, Africallagma vaginale, Agriocnemis forcipata, Ceriagrion annulatum, Pseudagrion kibalense, Anax congoliath, Neurogomphus alius, Paragomphus machadoi, Phyllomacromia aureozona, Acisoma trifidum, Cyanothemis simpsoni, Hadrothemis camarensis, Malgassophlebia bispina, Micromacromia camerunica, Neodythemis klingi, Notiothemis robertsi, Orthetrum austeni, Oxythemis phoenicosceles, Porpax asperipes, Thermochoria equivocata, Trithemis basitincta and Zygonyx regisalberti. Nine species confined to the Lower Guinea, the forest area that stretches between the Congo Basin and Atlantic Ocean from Cameroon to Gabon and western Congo, reach their southern limit in north-western Angola: Sapho orichalcea, Chlorocypha cyanifrons, Platycypha rufitibia, Copera congolensis, Pseudagrion simonae, Paragomphus abnormis, P. cf. darwalli and Tetrathemis fraseri.

Nearly 20% favour the swamps, grasslands, miombo woodlands and gallery forests that stretch eastwards, mostly above 1,000 m asl. This fauna is concentrated in Katanga and northern Zambia, but has now been proven to extend across to the Angolan upland with species like Umma electa, Chlorocypha fabamacula, Aciagrion steeleae, Ceriagrion sakejii, Pinheyagrion angolicum, Pseudagrion greeni, P. inconspicuum, P. deningi, Phyllomacromia unifasciata, Aethiothemis ellioti, Crocothemis brevistigma, Eleuthemis libera, Nesciothemis fitzgeraldi, Orthetrum macrostigma, Porpax risi, Rhyothemis mariposa, Trithemis anomala, T. leakeyi and Zygonyx denticulatus. The discovery of Trithemis integra near Uíge is of special interest, as it seemed endemic to the Albertine Rift, being known previously only from western Tanzania and Uganda and eastern DRC.

Sixteen species have so far only been found in Angola. With the exception of two known only from their type locality, all are limited to the central plateau (Figs 3, 5): the type locality for Platycypha rubriventris is questionable (see comment above) as may that of Pseudagrion dundoense from the extreme north-east, which could also be a river species from the very poorly sampled southern Congo Basin. No endemics have been found below 1,200 m asl in the east, although some drop down to about 500 m asl west of the escarpment. While the proportion of endemics (7%) is lower than for Ethiopia (12 endemics; 11%) and South Africa (30 endemics; 18%), countries that also enclose distinct highland areas, this still ranks Angola as one of Africa’s greatest centres of endemism for Odonata, rivalling the highlands of Cameroon (13 endemics) and the Albertine Rift, Eastern Arc and Katanga. Moreover, the number stands to increase, as almost two-fifths were described since exploration was reinitiated and at least another probable Platycypha species is already known to us.

Only Platycypha presents an endemic radiation. While the genus Chlorocypha has diversified with almost 30 species largely in the forested lowlands of west and central Africa, Platycypha is more varied and includes at least four open-land species (Fig. 4): P. caligata occurs widely from South Africa to Ethiopia, two species are limited to cooler parts of South Africa and Zimbabwe, one to Lake Tanganyika’s shores, and another possibly to Lake Malawi. Two species of submontane forest are endemic to Tanzania and Kenya. Four species are mostly in forest, three in the Lower Guinea and Congo Basin, while P. lacustris stretches around that basin and is thus expected in Angola too. The Angolan endemics are found mainly between 1,300 and 1,800 m asl in open habitats. Such a local radiation of a group that has otherwise diversified in the highlands to the east, and forests to the north, fits the overall affinities of Angola’s endemic Odonata both geographically and ecologically.

The four endemic Pseudagrion species, for example, have separate origins but similar links: morphology and COI data suggest that the nearest relatives of P. angolense and P. estesi are the rainforest species P. grilloti Legrand, 1987 and P. kibalense respectively. The former is limited to Congo and Gabon but the latter extends to Cameroon and Uganda. P. sarepi is closely related to P. fisheri and P. greeni, both of which extend from Angola into Zambia. While these species belong to the genus’s A-group, the B-group species P. dundoense is known only from Dundo and (as noted above) may not be endemic at all. Notogomphus kimpavita is the sister-species of N. praetorius found in highlands across southern Africa (including Angola), while Eleuthemis eogaster is nearest to an unnamed species from Gabon (Dijkstra et al. 2015). Molecular data for Umma femina and Onychogomphus rossii are lacking, but their morphology is close to U. electa and pale Onychogomphus species, both from the open plateaus stretching from Angola to Zambia and Katanga.

Thus, like the majority of Angola’s Odonata, most endemics probably originated quite recently and proximally from the forests to the north and open habitats to the east. However, some affinities are unresolved and potentially more distant: Agriocnemis toto and especially A. canuango have no obvious close relatives (Dijkstra et al. 2015), while the near-endemic A. angolensis and A. bumhilli are probably related to each other but even more distinct overall (Dijkstra unpubl.). Indeed, Angola may be the centre of diversification of Africa’s smallest damselflies. Micromacromia flava (Fig. 6b) is morphologically nearest M. miraculosa (Förster, 1906), known only from the East Usambara Mountains of north-eastern Tanzania (Dijkstra and Vick 2006), and the only one of four Micromacromia species adapted to non-forest habitats, being strongly pruinose with maturity. Elattoneura tarbotonorum may be closest to E. frenulata of south-western South Africa (Dijkstra et al. 2015).

If we compare the tallies for the well-studied neighbours Zambia and Namibia, the total number of species in Angola should lie somewhat above 300, meaning that less than 80% of the fauna is currently known. Additions can be expected throughout the country, but especially on the eastern and particularly northern border. Appropriately, the province of Lunda Norte should be the richest area of discovery, around Dundo where exploration began in the 1950s. On the other hand, the central highlands could expect more suprises, like the discovery of additional endemic species, with three areas being especially notable.

Firstly, despite having most records, the north-south directed range that lies entirely above 1,600 m asl and includes the Serra do Chilengue, Serra da Chela and Angola’s highest peak at Mt Moco (2,620 m asl) is poorly sampled as the large gaps in Fig. 1 and 3 illustrate. Secondly, except for at its extreme northern and southern ends, the western escarpment has not been surveyed at all. Finally, a huge plateau at 1,200-1,600 m asl stretches east from the Bié Highlands. Except for its southern edge, this area shared between Bié and Moxico Provinces, which is almost as large as Uganda (or the United Kingdom), has no records. These deep Kalahari sands are the ‘watertower’ of Angola and its neighbours, incorporating the headwaters of the Cuito, Cuando, Chicapa, Cuango, Cuanza and large tributaries of the Congo and Zambezi like the Cassai and Lungué-Bungo. The vast catchments of the Congo, Cuanza, Okavango and Zambezi meet in small area between Munhango and Cangonga. Watersheds are prone to endemism (Dijkstra et al. 2011) and this region is the top priority for further research.

New species are most likely to be found among genera prone to narrow (highland) ranges, i.e. with known Angolan endemics like Platycypha and Pseudagrion, but also Agriocnemis, Elattoneura, Notogomphus and perhaps Paragomphus. However, given the biogeographic diversity of Angola’s fauna and endemics, we could expect greater surprises. Among forest genera with no known Angolan endemic, Allocnemis seems most likely to reveal one, e.g. on the escarpment. The presence (or local endemism) of distinctly Lower Guinean genera like Neurolestes, Africocypha, Pentaphlebia and Stenocnemis seems less likely, but the Lower Guinean Stenocypha gracilis (Karsch, 1899) has four endemic relatives in the Albertine Rift and the sister-taxon of the Upper and Lower Guinean Tragogomphus is Nepogomphoides stuhlmanni (Karsch, 1899) in the Eastern Arc, suggesting an Angolan taxon is possible.

Three typical African highland genera are notably absent from Angola. Atoconeura is most likely to be present, being found in Zambia, Katanga, the Lower Guinea and Albertine Rift. However, its absence also from South Africa suggests historical factors may have been limiting, e.g. that the highlands were uninhabitable in cooler periods and unreachable when habitats were suitable (Dijkstra 2006). This might not apply to Proischnura, present in South Africa as well as Cameroon and the Albertine Rift. However, that genus is absent from Katanga and northern Zambia, which lies lower and thus possibly provided no stepping-stone to the mountains of Angola. Similarly, Zosteraeschna extends from the Cape to the Albertine Rift, Ethiopia and even northern Namibia, but is absent from Katanga, northern Zambia and also Cameroon, potentially providing no route to Angola.

If these genera do occur in Angola, they might only occur above 1,400 m asl and could constitute endemic species. More probable, however, is the discovery of endemics in genera that are well-represented across the country and continent, and that have highland endemics elsewhere but not in Angola, such as Africallagma, Neodythemis and Orthetrum. The only record of Pinheyschna rileyi in upland Angola is of a female, which might also turn out to represent a distinctive population. Finally, there is a remote chance that a (mainly) South African genus like Syncordulia or Ceratogomphus is present. As the family Synlestidae is also represented by Nubiolestes in the Lower Guinea, while Chlorolestes is also probable.