Species: Grammomys surdaster (Thomas & Wroughton, 1908)

Synonymy:

Thamnomys surdaster Thomas & Wroughton, 1908

Thamnomys oblitus Osgood, 1910

Thamnomys usambarae Matschie, 1915

Grammonys surdaster angolensis Hill & Carter, 1937

Grammomys vumbaensis Roberts, 1938

Grammomys caniceps Hutterer & Dieterlen, 1984

Type locality: Malawi, Zomba .

Distribution: Widely distributed from south-western Democratic RepublicoftheCongo(Kikwitarea) andAngolanescarpment,through Zambia to the Albertine Rift Mountains, Eastern Arc Mountains, and Southern Rift Mountains. It is possible that isolated marginal populations occur south of the Zambezi River, such as in Gorongoza National Park in Mozambique or at the border between Mozambique and Zimbabwe (typelocalityof vumbaensis),andincoastalforestsinKenya and Somalia (previously described as caniceps).

Comments: The species has frequently been reported as G. dolichurus in previous studies (Musser and Carleton 2005); however, G. dolichurus is restricted to South Africa (see above). The type of usambarae from northern Tanzania was synonymized with G. macmillani (Musser and Carleton 2005), but no representatives of the macmillani group are distributed in the Usambara Mountains, from where usambarae was described. We genotyped numerous specimens reported by Stanley and Goodman (2011), and they represent either G. cometes (generally larger body size) or G. surdaster (smaller body size). Although we were not able to include the type of usambarae in our morphological analysis, based on the smaller body size and short molar rows, we synonymize usambarae with G. surdaster (see similar opinion by Smithers and Wilson 1979 and Meester et al. 1986, who listed them under G. dolichurus). The populations in more arid coastal forests in Kenya and Somalia have smaller body size and lighter coloration and were described as G. caniceps (Hutterer and Dieterlen 1984) . Their mtDNA belongs unequivocally to the surdaster clade (we sequenced several specimens from the type series), as sister to su11 from the Lake Manyara National Park in northern Tanzania, but unfortunately, we failed to obtain genomic ddRAD data from the type series of G. caniceps because of low DNA quality. The skulls of the holotype and most paratypes of caniceps are classified with the surdaster group, as is the holotype of oblitus from a geographically close Voi locality (Table 1). The type of angolensis was classified into the surdaster group with PP = 0.73, and G. surdaster is the only species documented genetically in Angola (Fig. 6). The type of vumbaensis from eastern Zimbabwe (Vumba and Mount Selinda) was synonymized with G. macmillani (Musser and Carleton 2005), but no representatives of the macmillani group (sensu Bryja et al. 2017) are distributed in this region (see Figs 6, 7). Instead, it might represent the southernmost extension of the range of G. surdaster, because it was classified to this species with PP = 0.84.

Karyotype: Two specimens of G. surdaster from Uluguru Mountains had karyotype 2 n = 42, NFa = 64 (Corti et al. 2005); we sequenced both karyotyped specimens, and they belong to the lineage su10. This karyotype is likely to differ from that of G. caniceps (2 n = 56, NFa = 64; based on the figure of the karyotype presented by Hutterer and Dieterlen 1984) only by several Robertsonian translocations. The karyotypes of specimens from Somalia (Roche et al. 1984) are variable between and even within individuals, which is caused by polymorphism in Robertsonian translocations and the presence of numerous B chromosomes. However, the number of larger autosomal arms (unlikely to be B chromosomes) is again very similar or even identical to the karyotype of G. surdaster from the Uluguru Mountains (Corti et al. 2005). The karyotypes from Zambia (2 n = 50, NFa = 61–62, mtDNA lineage su7; Corti et al. 2005) and the Katanga region in Democratic Republic of the Congo (2 n = 52, NFa = 62; Matthey 1971, Petter and Tranier 1975) are similar, with slightly lower NFa, which can be explained by a single pericentric inversion.