Grammomys macmillani ( Wroughton, 1907 )

Species: Grammomys macmillani ( Wroughton, 1907) View in CoL

Synonymy:

Thamnomys macmillani Wroughton, 1907

Thamnomys discolor Thomas, 1910

Thamnomys surdaster elgonis Thomas, 1910

Thamnomys macmillani gazellae Thomas, 1910

Thamnomys surdaster insignis Dollman, 1911

Thamnomys buntingi Thomas, 1911

Grammomys aridulus Thomas and Hinton, 1923

Thamnomys (Grammomys) surdaster callithrix Hatt, 1934

Grammomys macmillani erythropygus Setzer, 1956

Type locality: Ethiopia, north of Lake Rudolf (= Turkana), Wouida, elevation 6200 feet (1890 m a.s.l.) .

Distribution: The belt of a mosaic of Guineo-Congolian forests and Sudanian savannah, from Sierra Leone to south-western Ethiopia and western Kenya ( Fig. 7 View Figure 7 ). In the central parts of the Sudanian savannah, it has been reported as G. dolichurus or G. macmillani (e.g. GBIF database;https://www.gbif.org/), but it is not confirmed genetically. In Uganda, it is the most widespread species, recorded even in Bugala Island in Lake Victoria ( Fig. 7 View Figure 7 ).

Comments: This species corresponds to INFOMAP MOTU VI, encompassing West African mtDNA clade m3 (formerly called buntingi ) and Central and East African mtDNA clade m5 ( Fig. 3 View Figure 3 ). The two clades are sister in both mtDNA and ddRAD phylogenetic analysis; in our view, they simply represent intraspecific (phylogeographical) genetic variation. A similar pattern of west–east differentiation has been found in numerous rodent taxa living at the northern margin of Guineo-Congolian forests and in Sudanian savannah (e.g. Mus musculoides, Temminck, 1853 , Bryja et al. 2014; Mastomys erythroleucus (Temminck, 1853) , Brouat et al. 2009; Arvicanthis niloticus (Desmarest, 1822) , Bryja et al. 2019a; Lemniscomys zebra (Heuglin, 1864) , Hánová et al. 2021; and Gerbilliscus giffardi (Wroughton, 1906) , Granjon et al. 2012).

There are many names previously used for Grammomys in this geographical range ( Fig. 7 View Figure 7 ). The oldest name is macmillani , described from Wouida, north of Lake Rudolf (= Turkana). This area can be inhabited also by the mtDNA clade m4 (INFOMAP MOTU IV, found in well-grown forests in higher elevations, from the Kenyan highlands to south-western Ethiopia, considered here as G. ibeanus ; see below). However, the macmillani type instead represents the widespread species in the belt of Sudanian savannah. Firstly, the type locality is not described precisely. Based on the species description ( Wroughton 1907), the holotype was collected by the expedition of Mr Ph. C. Zaphiro on 30 June 1905 in Wouida, north of Lake Rudolf, elevation 6200 ft (1890 m a.s.l.). We were unable to identify habitats in ‘Wouida’, and we did not capture any specimens with m5 mtDNA in southern Ethiopia, but the specimen from this mitochondrial clade from the same elevation in South Sudan clustered unequivocally with MOTU VI in all ddRAD analyses. Many other taxa typical of Sudanian savannah were recently documented from drier woodlands in south-western Ethiopia (e.g. Gerbilliscus giffardi or Aethomys hindei (Thomas, 1902) ; Bryja et al. 2019b) and north-western Uganda (S.W.Babyesiza, J. Bryja, pers. observation). The presence of this Grammomys clade is therefore very likely in drier woodlands in the region ‘north of Lake Rudolf ’, i.e. the type locality of G. macmillani , which also represents its easternmost record. Second, two specimens from Chak Chak (now South Sudan) are mentioned to be almost identical with the type of macmillani from Wouida in the description by Wroughton (1907). They were later described as G. macmillani gazellae by Thomas (1910), but the investigation by Hutterer and Dieterlen (1984) again confirmed that the types of macmillani and gazellae are virtually identical.

Spermatozoal morphology was documented by Breed (1995; as gazellae ). The reports of G. macmillani in the Eastern Arc Mountains of Tanzania, documented by Stanley et al. (1998), represent another taxon (most likely MOTU VII, for which we propose the name G. polionops ; see below).

Setzer (1956) suggested (without any genetic data) that none of the characters formerly used to separate G. macmillani and G. surdaster is of more than subspecific value, and he proposed that the animals formerly known as surdaster should henceforth be called macmillani . Our morphological analysis suggests high similarity of these taxa ( Fig. 4 View Figure 4 ; Table 1 View Table 1 ), but they are distinct both genetically and geographically. Based on the locations of type localities and distributions of the genomically well-delimited surdaster and macmillani groups (sensu Bryja et al. 2017), we agree with Musser and Carleton (2005) and place in synonymy with G. macmillani the following names: callithrix, gazellae , and erythropygus. Furthermore, discolor , elgonis, and insignis were synonymized with G. dolichurus ( = surdaster in the work of Setzer 1956) but should also be here listed as synonyms of G. macmillani , based on the locations of their type localities.

The taxon aridulus was listed either as a subspecies of G. macmillani ( Allen 1939, Ellerman 1941, Setzer 1956) or included in G. dolichurus ( Misonne 1974) but was considered a distinct species by Hutterer and Dieterlen (1984). Based on its description, it represents ‘an unusually pallid, desert-coloured species … skull resembles G. buntingi from Liberia … In outward appearance it is perhaps more like G. s. elgonis than any other, while in skull it resembles the outwardly very different G. buntingi .’. We obtained a short CYTB sequence from two specimens of the type series collected on Mt. Jebel Marra in Sudan, and they clearly cluster with m5. We therefore propose to synonymize aridulus with G. macmillani , which also makes sense from a biogeographical perspective ( Fig. 7 View Figure 7 ).

Except for Misonne (1974), who included it in G. dolichurus , buntingi has always been listed or discussed as a distinct species in all recent reviews (Petter and Trainer 1975, Hutterer and Dieterlen 1984, Musser and Carleton 2005, Kryštufek 2008, Denys et al. 2011). The western African population forms a separate subclade in both mtDNA ( Fig. 2 View Figure 2 ) and ddRAD ( Fig. 3 View Figure 3 ) phylogenies, but INFOMAP clustered them together with central and eastern African populations into MOTU VI ( Fig. 3 View Figure 3 ). Morphologically, these populations are very similar, differing by very small details of the teeth ( Petter and Tranier 1975), and we were not able to find any clearly distinguishing traits compared with populations of Grammomys from western Africa and the Central African Republic in the MNHN collection. Based on the genetic and morphological similarity, we therefore propose to synonymize buntingi with G. macmillani and consider it as an intraspecific phylogeographical lineage.

Karyotype: Karyotype was analysed by Petter and Tranier (1975; as buntingi and gazellae ) and Civitelli et al. (1989; as gazellae ). The population of buntingi from Cote d’Ivoire has 2 n = 52, NFa = 58 (Peter and Tranier 1975). The karyotype of gazellae from Central African Republic is highly variable because of the presence of numerous B chromosomes, but if they are excluded, then 2 n = 54, NFa = 58 (based on fig. 2 of Civitelli et al. 1989), i.e. they differ by only a single Robertsonian translocation, supporting their conspecificity.