The genus Ectonura Cassagnau, 1980in South Africa (Collembola, Neanuridae, Neanurinae), with a key to South African Neanurinae

Abstract Two new species of Neanurinae (Collembola) are described from the Western Cape, South Africa: Ectonura monochaeta sp. n. and Ectonura barrai sp. n. Ectonura monochaeta sp. n. differs from other species in the genus by its strongly reduced chaetotaxy, and the lateral shift of dorso-internal chaetae on Abd. V and their integration in the tubercles (De+DL). Ectonura barrai sp. n. is similar to Ectonura natalensis (Womersley, 1934), but differs in chaetotaxic details and chaetal group arrangement. A key to the seven species of Neanurinae recorded from South Africa is given.

introduction Neanurinae Collembola are represented in tropical Africa by a large number of species in the tribe Paleonurini, and a single representative of the tribe Neanurini, the parthenogenetic species Neanura muscorum (Templeton, 1835). However, only a few areas have been sampled outside the mountain ranges of Eastern Africa (Cassagnau 1991(Cassagnau , 1996(Cassagnau , 2000Weiner and Najt 1998). In South Africa, only three genera and five species of Neanurinae have been recorded so far ( Fig. 1): Neanura MacGillivray, 1893 with N. muscorum;Vitronura Yosii, 1969 with V. joanna (Coates, 1968), and Ectonura Cassagnau, 1980 with E. natalensis (Womersley, 1934), E. oribiensis (Coates, 1968) and E. coatesi Barra, 1994. N. muscorum has been probably introduced from Europe, where all other species of the genus Neanura occur. The genus Vitronura, diversified in China, Sunda and western Pacific islands, has a single widespread species, V. giselae (Gisin, 1950), that occurs both in the tropics and in gardens in Europe. Therefore, the presence of another distinct species isolated in South Africa requires confirmation, as V. joanna is morphologically very close to V. giselae. The genus Ectonura, limited to South Africa in the African continent, includes the only Neanurinae unambiguously native of South Africa; the genus is otherwise present in New Caledonia with 11 species (Deharveng and Bedos 2002), and an undescribed species is recorded from South Australia by Greenslade and Deharveng (1991).
Among the large amount of samples recently collected in the Western Cape Province in the frame of the Franco-South African PROTEA project "Uncovering Springtail Diversity in the South African Cape Floristic Region: a combined taxonomic and barcoding approach", we retrieved representatives of the three cited genera, including new Ectonura species, as well as a single species of a fourth genus, Paleonura Cassagnau, 1982. This confirms that South Africa fauna of Neanurinae is particularly poor, com- pared to that of East African mountains or Madagascar (Cassagnau 1996(Cassagnau , 2000, or other gondwanian territories such as Australia (Greenslade and Deharveng 1991). Its richness in Neanurinae is actually similar to that of southern America subtemperate areas, where N. muscorum also occurs, together with a few endemic Paleonura (Cassagnau and Oliveira 1990).
Ectonura is therefore the most diversified genus of Neanurinae in southern Africa. Several undescribed species were present in our samples, mostly as isolated specimens. Two of them were collected in sufficient number and are described in this paper: Ectonura monochaeta sp. n. from Table Mountain and Ectonura barrai sp. n. from Grootvadersbosch, both located in the Western Cape Province.

Head chaetotaxy as in
Tergite chaetotaxy as in Table 1B and Fig. 2A. Chaeta Di absent on Th. I. Tubercles De and DL separate on Abd. IV. Tubercle L of Abd. IV shift ahead the tubercle line Di-De-DL. Tubercles Di, De and DL fused on Abd. V on each side of axis. Tubercle Di of Abd. V with Di1 macrochaeta, Di2 and Di3 absent. Abd. VI not or hardly bilobed, with strong secondary granules, present even on the axis. S-chaetotaxic formula: 2+ms, 2/11111. Ventral chaetotaxy similar to that of E. barrai sp. n., except the furcal rest in some specimens (Fig. 4C). Secondary sexual characters well developed in the adult male consists of chaetae Ag1 and Ag2 of Abd. V strongly thickened and serrated (Figs 2E, 3F), and chaetae of furcal rest, some Ve of Abd. IV (Fig. 3D), sometimes Ag3 of Abd. V (Fig. 3F), and 3+3 Ve of Abd. VI serrated but less strongly. In a male juvenile from Jonkershoek, chaetae Ag1 were bifid (Fig. 3E).
Microchaetae of furcal rest smaller than secondary granules, often unconspicuous (Fig. 2D). Leg chaetotaxy given in Table 1C. Tita without chaeta M and with chaetae B4-B5 short, not longer than other long chaetae of Tita (Fig. 2C). Claw untoothed, not striated in its basal part, and devoid of secondary granulation.
Derivatio nominis. The species name refers to its reduced chaetotaxy of dorsointernal tubercles of tergites, which bear only one chaeta from Th. II to Abd. IV (2 or 3 in other species of the genus).
Ecology. All known localities of Ectonura monochaeta sp. n. belong to the Southern Afrotemperate Forest vegetation type. The species is common in this habitat, typically found in the Western Cape, but absent in shrub formations of the fynbos. The distribution ranges from Table Mountain National Park to Jonkershoek Nature Reserve, Stellenbosch. The species is mixed in Stellenbosch with another undescribed species of Ectonura.
Discussion. The new species Ectonura monochaeta is unique in the genus by the lateral shift of dorso-internal chaetae on Abd. V and their integration in the tubercles (De+DL). Such a lateral shift is only know in Ectonura paralata Deharveng, Weiner & Najt, 1997 from New Caledonia, but less marked and without integration of Di chaetae in (De+DL). By other chaetotaxic characters (Di2 and De2 present on head and on tergites of Th. II-Abd. IV; D, E, OcA present on head) and tubercle arrangement (tubercle Di not developed, others as large flat plates), E. paralata is however only remotely related to our species. E. monochaeta is also distinct from other species of the genus Ectonura by the strong reduction of its chaetotaxy: absence of several chaetae on head (A, O, C, D, E, Oca), absence of Di2, De2 and DL2 on tergites, only 2+2 dorsal chaetae on Th. I, and only 6+6 chaetae on Abd. VI.
The lateral shift of Di on Abd. V is one of the characteristic feature of two genera, the monotypic genus Zelandanura Deharveng &Wise, 1991 from Campbell Island andPronura Delamare Debouteville, 1953 which is highly diversified in Africa and in Asia. Zelandanura differs from Ectonura by the fusion in one plate of all tubercles of the central area of head, and by the fusion of Di tubercles on the axis on Abd. IV. Contrary to Ectonura, chaetae of the central area of head are not separated in two groups on both side of the axis in Pronura.
Derivatio nominis. This species is named in honour of Jean-Auguste Barra, for his important contribution to the knowledge of South African Collembola.
The genus Ectonura Cassagnau, 1980 in South Africa (Collembola, Neanuridae, Neanurinae)... 43 Ecology. This species was collected in the yellowwood forest leaf litter of Grootvadersbosch Nature Reserve. This is a remnant forest of the larger Tsitsikamma Forest Reserve situated 300 km to the south. The forest consists of indigenous trees such as yellowwood, ironwood and stinkwood.
Discussion. Ectonura barrai sp. n. is similar to E. natalensis in its relatively complete chaetotaxy, but different in several details of chaetal arrangement. Based on the redescription of Coates (1968), E. natalensis has a more reduced chaetotaxy than E. barrai sp. n.: absence of meso/microchaetae D, Di2 and De2 on head, absence of De2 on Th. I, absence of De3 and DL3 on Th. II-III, absence of Di2 on Abd IV-V. Conversely, chaetae Di2, De2 and DL2 on Abd. I-III of E. natalensis are much larger than homologous chaetae of E. barrai sp. n. (macrochaetae or large mesochaetae versus short mesochaetae). Chaetal groups De and DL of Abd. IV are separate in E. natalensis versus fused in E. barrai sp. n., and chaetal groups Di of Abd. V are fused on axis in E. natalensis versus separate in E. barrai sp. n. The unusual arrangement of S-chaetae on Ant. IV figured by Coates (1968) is probably an erroneous interpretation, and not considered here as a valid differential character.