Grammomys cometes ( Thomas & Wroughton, 1908 )

Species: Grammomys cometes ( Thomas & Wroughton, 1908) View in CoL

Synonymy:

Thamnomys cometes Thomas & Wroughton, 1908

Thamnomys dolichurus littoralis Heller, 1912

Grammomys silindensis Roberts, 1938

Grammomys selousi Denys et al., 2011

Type locality: Mozambique, Inhambane .

Distribution: Musser and Carleton (2005) specified the distribution of South African G. cometes as follows: the savanna woodland biome in southern Africa from Pirie Forest (north-west of King William’s Town) in south-eastern Eastern Cape Province of South Africa north through KwaZulu-Natal and Limpopo provinces into eastern Zimbabwe and Mozambique south of the Zambezi River (see references provided by Musser and Carleton 2005). We consider all members of the cometes group (sensu Bryja et al. 2017) as belonging to this species, hence its distribution is wider and also includes populations in the Southern Rift Mountains (including Mt. Gorongosa and Mt. Namuli in Mozambique, Mt. Mulanje in Malawi, and the Chimanimani Mountains on the Zimbabwe – Mozambique border), Eastern Arc Mountains, and coastal forests of Tanzania (including Mafia Island) and southern Kenya. However, we have no genetically confirmed records of this species in South Africa outside Kwazulu-Natal ( Fig. 5 View Figure 5 ), and its occurrence in the Eastern Cape province (the Pirie Forest specimens in the collection of AMNH) should be re-analysed genetically.

Comments: This is, on average, the largest species of the genus (Supporting Information, File S3). Musser and Carleton (2005) studied the holotype of G. cometes and the other specimens in the type series noted by Thomas and Wroughton (1908); these animals are, on average, larger and have more inflated bullae than those from north of the Zambezi River (but these might represent Grammomys surdaster , as considered here; see below). Ansell (1978) and Ansell and Dowsett (1988) assigned samples from Zambia and Malawi to G. cometes but were also impressed by the chromatic and morphological contrast between them and the holotype from Inhambane. The population from the Pirie Forest was discussed by Taylor et al. (1994) and Taylor (1998), who noted sympatry between G. cometes and G. dolichurus at some localities and at others a gradation of the diagnostic traits usually used to distinguish the two species. Whether those characters have limited discriminatory use in these areas or whether hybridization is occurring is unclear; however, the two species are clearly distinguishable by all genetic markers in our study. We observed that G. cometes has no axial and four inguinal teats (a trait shared with G. dryas , suggesting its ancestral character). This was already used as a discriminating traits by Roberts (1938), who described silindensis (cometes group) and vumbaensis (surdaster group) with differing teat counts: 0 + 4 in the former and 2 + 4 in the latter. The genotyped specimens from the Southern Rift Mountains (especially Malawi) and Eastern Arc Mountains (e.g. Usambara) were reported as G. ibeanus in previous studies (e.g. Stanley and Goodman 2011) and collections (mainly FMNH), but G. ibeanus is not distributed in these mountains (see its account).

The short DNA barcode of the type of silindensis belongs to the se2 lineage, and the holotype and paratype of selousi have mtDNA of se4 lineage (Supporting Information, File S1). We therefore propose to synonymize both of them with G. cometes . All available types bearing the names silindensis , selousi , and cometes cluster unequivocally with the genotyped specimens from the cometes group in the geometric morphometric analysis ( Table 1 View Table 1 ). Intriguingly, the type of littoralis was classified as G. cometes , which might represent an extension of its documented distribution in the coastal forests of southern Kenya. The species represents a biogeographical element typical for the coastal forests of eastern Africa (sensu Fayolle et al. 2014).

Phylogenomic divergence dating analysis by Mikula et al. (2021) suggested the split of the cometes group (i.e. G. cometes ) from the rest of the genus in the Pliocene (3.2 Mya), which agrees with the opinion that most of the coastal forest endemics, including mammals, are palaeoendemics ( Burgess et al. 1998). The north–south structuring of the species in mtDNA reflects the fragmented nature of coastal forests, possibly maintained by large rivers flowing to the Indian Ocean (e.g. Rufiji, Zambezi, and Limpopo) or repeated shrinking of coastal forests and their shift to higher elevation owing to climate changes and increases in sea level ( Burgess et al. 1998). This suggests that species living in coastal forests colonized parts of the Eastern Arc Mountains and Southern Rift Mountains, probably via riverine gallery forests (see also J. Krásová, O. Mikula, J. Bryja, R. Šumbera, unpublished observations).

Karyotype (see Fig.2 View Figure 2 ): 2 n = 49–50, FNa (autosomal fundamental number) = 56, FN (fundamental number) = 58 ( Denys et al. 2011, as G. selousi ). The karyotypes reported as G. dolichurus (2 n = 44; Woodbush Forest Reserve, South Africa; Dippenaar et al. 1983) and G. ibeanus (2 n = 44–48; Nyika Plateau, Malawi; Chitaukali et al. 2001) might represent G. cometes . All specimens reported as G. cometes (2 n = 52) by Kryštufek et al. (2008) are G. dolichurus (see Bryja et al. 2017 for more details).