Dendromys lovati de Winton, 1900

Mulualem, Getachew, Lavrenchenko, Leonid A., Kerbis Peterhans, Julian C., Mikula, Ondřej, Bryjová, Anna, Meheretu, Yonas & Bryja, Josef, 2024, Taxonomic revision and evolutionary history of the climbing mice in eastern Africa (Dendromus mystacalis clade): the role of elevation and geographical barriers in the speciation process, Organisms Diversity & Evolution 24 (4), pp. 573-598 : 586-594

Dendromys lovati de Winton, 1900

Holotype: BMNH 3.3 .21

Type locality: Ethiopia, Managasha (Menangesha), 7000 ft. (2134 m a.s.l.), picked up on grass plateau, near Addis Ababa

Taxonomic notes: The species was previously placed in its own subgenus, Chortomys , but the first molecular genetic study by Lavrenchenko et al. (2017) showed that it belongs to Dendromus sensu stricto. It is the only species of Dendromus with four mid-dorsal stripes, and it is a sister species to all other taxa of the Ethiopian clade. Other specific traits, lineatus (c). Four remaining species in the D. mystacalis complex: non-genotyped individuals of D. “mystacalis ” (sensu Voelker et al., 2021), described as a new species, D. pseudomystacalis here, and D. sp. indet. 2 (which remains unresolved and could represent a new species or be conspecific with D. pseudomystacalis ) are represented by white circles; data are taken from Yalden et al. (1976), Dieterlen (2009), and Meheretu et al. (2022) (d). For the genotyped material, the colours are as follows: red = D. ruppi ; pink = D. sp. indet. 4 (described as a new species, D. rogersi here); orange = D. sp. indet. 2; green = D. “mystacalis ” (sensu Voelker et al., 2021)

such as its long and soft fur, short tail, and large and rounded ears, are considered adaptations to a terrestrial habitus amid Afroalpine grasslands and heathlands. The population on Mt. Choqa is very distinct genetically ( Kostin et al., 2020, this study), but, awaiting further studies, we consider it conspecific with D. lovati .

Ecological notes: The species is nocturnal ( Meheretu et al., 2022). Unlike other species of Dendromus , it lives mostly on the ground ( Yalden et al., 1976) and builds a grass nest under boulders ( Wilson et al., 2017). We captured only D. “mystacalis ” (sensu Voelker et al., 2021) in the type locality of D. lovati (plateau on Mt. Menangesha, 2880 m a.s.l.) in 2012. The whole locality was overgrazed, and it is possible that D. lovati , which seems dependent on high grasslands, is extinct there. The same two species were genetically confirmed in the Semien Mts. ( Craig et al., 2020). They live there in sympatry at 3250 m a.s.l., but D. lovati generally prefers higher elevation and was the only Dendromus captured above 3600 m a.s.l.

Distribution: The Afroalpine grasslands and heathlands above 2500 m a.s.l. on both sides of the Great Rift Valley ( Fig. 9a ). The recent records from the Semien Mts. at 4000 m a.s.l. ( Craig et al., 2020) and Mt. Choqa ( Kostin et al., 2020) (> 3900 m a.s.l.) represent the highest known localities of the species.

2) Dendromus abyssinicus Osgood (1936) resurrection of the species name

Dendromus insignis abyssinicus Osgood (1936)

Dendromus insignis abyssinicus ( Allen, 1939: list of taxa)

Dendromus mesomelas (part.) ( Misonne, 1971)

Dendromus (Dendromus) mesomelas (Yalden, et al., 1976)

Dendromus insignis ( Dieterlen, 2009: new records)

Dendromus insignis (Carleton & Musser, 2005)

Dendromus insignis ( Dieterlen, 2013)

Dendromus sp. (part.) (Lavrenchenko et al., 2017: phylogeny, karyotype)

Dendromus sp. (et al., 2019: new records, altitudinal distribution)

Dendromus sp. indet. 1 ( Bryja et al., 2019: distribution)

Dendromus sp. indet. 1 ( Voelker et al., 2021: phylogeny)

Holotype: Field Museum of Natural History , Division of Mammals, number FMNH28656 , field number 5993, collected by W. H. Osgood on November 9, 1926.

Type locality: Ethiopia, East slope of Mount Albasso (Mt. Badda), Chilalo Mountains, Arusi (Arsi), 11,000 ft. (3,353 m a.s.l.)

Taxonomic notes: This taxon was originally described as a small member of D. insignis ( Thomas, 1903; type locality: Nandi, Kenya). However, Voelker et al. (2021) were the first to show the Ethiopian populations are embedded within the D. mystacalis clade and sister (at mtDNA markers) to the uniquely four-striped D. lovati in a clade with mostly an Ethiopian distribution. Further collecting shows that this is a common taxon on both sides of the Rift Valley, endemic to the high elevations of Ethiopia. In our phylogenomic analysis, we included the topotypic specimen LAV2148 from the Arsi Mts. We also compared the morphology of our recently collected material on both sides of the Great Rift Valley with the holotype in the FMNH collection. Further, distinct chromosome diploid numbers (2n = 48 and 2n = 50, Lavrenchenko et al., 2017) in specimens on opposite sides of the Rift Valley and a considerable genetic distance between them (5.2% at the CYTB gene, Kostin et al., 2018) suggest that more than one taxon may be represented here. The grey roots to the belly fur, the single prominent dorsal stripe, sharply bicolour tail, and claw on the 5th hind toe make this species readily diagnosable. The hindfoot (18.6–20.5 mm) and tail (89–106 mm) are very long. The cranio-dental measurements are as follows: GLS 20.9–22.1 mm, CI 18.8–19.9 mm, HBC 7.2–7.9 mm, and UTR 3.2–3.6 mm.

Ecological notes: An Afroalpine species. Captured in habitats including grazed Hypericum , Erica shrubs, Helichrysum , and Erica next to juniper forest, regenerating Erica and grassland.

Distribution: Endemic to Ethiopia, it is found at high elevations on both sides of the Ethiopian Rift. On the Harar Plateau, it is known from Bale Mts. (northern slope of the Sanetti Plateau, 3750 m a.s.l., Sodota, 3507 m a.s.l.), Arsi Mts. (Mt. Albasso, 2800–3414 m a.s.l., Shirka area, 3294 m a.s.l., Chilalo Mt., 2799 m a.s.l.). On the Abyssinian Plateau, it is more widespread, with a more northerly population from Aboya Gara (3533 m a.s.l.) and Borena Sayent NP (3198 m a.s.l.) and a more southerly population from Guwasa (3320 m a.s.l.), Debra Sina (3233 m a.s.l.), and Ankober (3194 m a.s.l.). Three specimens in the BMNH (BMNH 72.1294, BMNH 72.1295, BMNH 76.113, leg. Largen et al. 1971) recorded from “Bale Dinshu” above 3000 m, as well as two further adult specimens in the SMNS from the same region (leg. Rupp and Nikolaus in 1971 and 1976, respectively; SMNS 23724 and SMNS 23983) belong to the same taxon ( Dieterlen, 2009). On the other hand, some of the lower elevation records referenced by Yalden et al. (1976) as D. mesomelas (Sheikh Mahomet, Addis Ababa, Alemaya = Haremaya) require confirmation and are not shown on the distribution map ( Fig. 9b ).

3) Dendromus mystacalis ( Heuglin, 1863) , revised description

Dendromys mystacalis Heuglin, 1863

Dendromus lineatus Heller, 1911 (Holotype: USNM164816 , adult male, collected by J. Alden Loring, February 1, 1910; field number 8921; Type locality: Rhino Camp , Lado Enclave, on the White Nile, Northwestern Uganda), part of the type series (but probably not the holotype) may belong to this species.

Dendromus mystacalis lineatus (part.) ( Dieterlen, 2009: distribution, new records)

Dendromus sp. (part) (Lavrenchenko et al., 2017: phylogeny, karyotype)

Dendromus sp. indet. 3 ( Bryja et al., 2019: distribution in Ethiopia)

Dendromus sp. indet. 3 ( Voelker et al., 2021: phylogeny)

Holotype: SMNS-Z-MAM-001055; a specimen collected by Heuglin in 1863 and sent to the Stuttgarter Naturalienkahiuett (now SMNS), where it was registered under no. 1055 in the (Accession) Catalogue for Mammals. The specimen was mounted in a climbing posture (see Dieterlen, 2009, Fig. 8 ).

Type locality: Eifag (Eifaz), north of Lake Tana at 12.03 N and 37.46 E, 2000 m a.s.l.

Taxonomic notes: The identity of D. mystacalis is crucial for any revisionary work of the genus as it is the oldest available name for the white-bellied group of Ethiopian Dendromus , a name which has subsequently been applied to all small and medium-bodied, white-bellied Dendromus throughout sub-Saharan Africa (e.g. Misonne, 1971; Carleton & Musser, 2005; Monadjem et al., 2015). Through ancient DNA analysis, we were able to reconstruct the mitogenome of the type specimen of D. mystacalis , it clusters with D. sp. indet. 3 (sensu Voelker et al., 2021). Since the type specimen of Dendromys mystacalis was described in Latin and in German in 1863, in a journal not readily available, we provide a translation of Heuglin (1863) for important passages. “ The type is an old adult male. Its chin and throat are a light red-ochre. We attribute the name mystacalis to its “moustache”, a reference to the white band of hairs across its upper lip. Its belly is pure white; the bases of its ears are a pale ochre; vibrissae have black dots; the nose, cheiridia, and nails are skin-coloured. The tail is long and finely haired, the length of the body 2.5″, and the length of the tail 3″. The colour of the dorsum and belly strongly contrast. Almost no sign of a grey dorsal stripe. Projecting large black eyes. The upper incisors clearly grooved. The nail of the hallux is very small. The liver contains 8 lobes, the kidneys are well-developed, the soft and muscular stomach is filled exclusively with plant material (perhaps juniper needles), the appendix is of medium size, the colon is 2″ 10/12″, and the small intestine is almost 8″ long. The type specimen was caught by hand after being disturbed while nesting within the nest of an eagle located at least 45′ high in a Juniperus tree, adjacent to a church next to the market in the town of Eifag “ ( Heuglin, 1863).

Dieterlen (2009) attempted to re-describe the taxon, but his observations of exclusively Ethiopian material include specimens that could pertain to two, perhaps three, species. Besides D. sp. indet. 3 (sensu Voelker et al., 2021;, i.e. true D. mystacalis ), these include D. “mystacalis ” (sensu Voelker et al., 2021, described below as a new species) and D. sp. indet. 2. Of relevance here, Dieterlen (2009) mentions (p. 196) three characters: (1) “The black mid-dorsal stripe is variable: in four specimens the stripe is complete from neck to root of tail, in three others the stripe is thin and not visible throughout its length, and in four there is no mid-dorsal stripe at all”. (2) “The hairs of the dorsal surface of the hind foot are somewhat reddish-brown, as are the short hairs on the ears”. (3) “The skin of the short-bristled tail is pigmented dark on the upper side and whitish below”. These traits possibly represent a mix of up to three taxa.

Here we sequenced the mitogenome of the holotype of D. mystacalis , and it unambiguously clustered with the lineage referred to as D. sp. indet. 3 by Voelker et al., (2021, Fig. 2 ). These specimens share the absence of a dorsal stripe ( Fig. 6 ) and occurrence in lower elevations of the Ethiopian highlands (the type locality of D. mystacalis at 2000 m a.s.l.). Also, the sequence of the holotype is sister to the sequences from the Dhati-Welel NP, which are the geographically closest confirmed localities to the type locality of D. mystacalis ( Fig. 9c ). Our review of specimens, genetically identified as D. mystacalis , enables us to provide the following re-description: the tail is generally unicolor (contra Dieterlen, 2009), and longer than the head and body (120 to 140%); the hind foot is 16–17.5 mm (with claw), and the adult weight is 9–11 g (Supplementary material S1); the small claw on pedal V; manual V with a small but distinct nub; white hairs (moustache) on the upper lip. There is no sub-auricular ear tuft. The ears are dark brown with orange hairs. The dorsum of the foot with white hairs (contra Dieterlen, 2009), and the dark patch on the ankle. The dorsal stripe is generally absent, but if present, very faint. Interestingly, when characterizing the holotype, Dieterlen (2009) already mentioned that “After 144 years of preservation, the reddish-brown dorsal pelage has become dull and the (originally?) weakly developed mid-dorsal stripe disappeared”. Despite this fact, he considered numerous specimens from the Ethiopian highlands with a well-developed dorsal black stripe conspecific with the holotype of D. mystacalis . The ventral pelage is white to roots, from chin, throat, and belly to vent). The type specimen has no skull, but the following cranio-dental measurements are available from genotyped specimens: GLS 19.8–20.8, CI 18.6–19.5, HBC 9.25–10, UTR 3.15–3.3. Chromosome number: 2n = 44 (Lavrenchenko et al., 2017).

Dieterlen (2009) analysed numerous material from lowland elevations of South Sudan and assigned them to D. mystacalis lineatus . We compared the genotyped specimens of D. sp. indet. 3 (sensu Voelker et al., 2021) from southwestern Ethiopia (Dhati-Welel and Sheko) and northeastern Uganda (Mt. Morungole) with the material from the Torit region in South Sudan (in FMNH Chicago), and we consider them conspecific. They are characterized by a purely white belly and a brownish or ochraceous dorsal part. The type series shows great variation in the distinctiveness of the black mid-dorsal stripe, from well-defined (e.g. in the holotype of D. lineatus ) to specimens that show only a very faint trace of it ( Dieterlen, 2009; Setzer, 1956). Most skins of the 17 specimens from South Sudan have the mid-dorsal stripe poorly developed, and in three specimens, the stripe is absent ( Dieterlen, 2009). The names D. acraeus and D. ruddi , mentioned by Dieterlen (2009) in relation to D. lineatus , most likely represent other species.

Ecological notes: The species is nocturnal and prefers grassland in relatively moist savannahs, but can nest high up in trees (bird nests) ( Dieterlen, 2009). It lives at the lowest elevation of all the species in the D. mystacalis clade. The elevational range of D. mystacalis lineatus in South Sudan varies between 630 and 1118 m a.s.l. ( Dieterlen, 2009), but in the foothills of the Ethiopian highlands it can live higher (1254–1600 m a.s.l., this study; type locality is at 2000 m a.s.l., Yalden et al., 1976), and in northeastern Uganda it was confirmed on Mt. Morungole at 2500 m a.s.l. (Supplementary material S1). Specimens from Mt. Morungole were all captured in high-elevation grassland. Specimen ETH1421 (Kite) was captured in a pitfall trap at the margin of the Sheko primary forest among shrubs, while ETH2644 (Bosa Wanche) was captured in a snap trap in the grassland of the Humbo Community Forest Area. Four specimens from Dhati-Welel were captured at the interface of wetlands with swamp grass vegetation ( Cyperus sp. ) and tall grass savannah with sparse Combretum-Terminalia woodland vegetation. When captured in late July/early August, two of the three males from Mt. Morungole had scrotal testes. The adult female from Bosa Wanche (October 31), had 4 + 1 embryos; the adult male from Kite (November 7) had scrotal testes; and 2 of 3 males captured on Mt. Morungole in late July/early August, had scrotal testes.

Distribution: The species is distributed in northeastern Uganda (Mt. Morungole), South Sudan (reviewed by Dieterlen, 2009, as D. mystacalis lineatus ), and southwestern Ethiopia (this study; Fig. 9c ). The southwestern limit of its distribution is not known. Material attributed to D. mystacalis lineatus in the West Nile region of NW Uganda ( Heller, 1911), the Kivu region or the Garamba National Park in the DRC (see Dieterlen, 2009) should be explored genetically. Genetically, it is confirmed from five mid-elevation localities in Ethiopia (the holotype from Yifag, 2000 m a.s.l. (“Eifag”, 11 km E of Lake Tana); Dhati-Welel National Park, 1443 m a.s.l.; Kite, Fanika Village, 1254 m a.s.l.; Ochollo, 1880 m a.s.l.; Bosa Wanche, Ambe Zaruwa Village, 1600 m a.s.l.); and one in the northeastern corner of Uganda (Mt Morungole, 2520 m a.s.l.). Dieterlen (2009) refers 11 additional specimens (BMNH, SMNS) from 8 additional localities in Ethiopia to D. mystacalis . Of special interest is BMNH1934.2.24.88 from Bab Bar Sir, 3 miles north of Lake Tana, just a few kilometres from the type locality. However, without genetic and morphological examination/confirmation, we cannot attribute any of these to D. mystacalis proper.

4) Dendromus pseudomystacalis Kerbis Peterhans, Lavrenchenko, Mulualem, Craig, Yonas & Bryja , sp. nov.

Dendromus mystacalis (part.) ( Yalden et al., 1976: distribution)

Dendromus melanotis (?) ( Bulatova et al., 1995: karyotype)

Dendromus melanotis ( Lavrenchenko et al., 2007)

Dendromus mystacalis (part.) ( Dieterlen, 2009: distribution)

Dendromus mystacalis ( Bryja et al., 2019: distribution)

Dendromus mystacalis (Craig, et al.,; 2020)

Dendromus mystacalis ( Voelker et al., 2021: phylogeny)

Holotype: Field Museum of Natural History, Division of Mammals number FMNH 229596 (field number EWC 12343), collected by E.W. Craig and Y. Meheretu, November 25, 2015. The type specimen, consisting of the skin, skull, and carcass in ethanol, is an adult female with fused basi-occipital suture and moderate dental wear of all three molars, teats 4 + 4. External measurements were made in the field: total length 158 mm, (head and body length 61 mm, infered); hindfoot length 18 mm (with claw); ear length 12 mm; weight 8.8 gr. Specimen caught in a pitfall.

Type locality: Ethiopia, Amhara Regional State, North Gondar Zone, Debark Woreda , Simien Mountains National Park , Sankober Camp : 13.23159° N, 38.03767° E, elevation 3240 m a.s.l.

Paratypes: Five additional specimens, all from the same locality and elevation at Sankober Camp, all FMNH: FMNH 229591 m , ssa; FMNH 229592 f, ssa ; FMNH 229594 m , ssa; FMNH 229811 m , asr; FMNH 229812 f, alc. After being returned to Ethiopia, the following specimens from the same locality were lost during the recent civil war (2020–2023) : FMNH 229597 f, ssa ; FMNH 229598 m , ssa; FMNH 229600 f, ssa ; FMNH 229601 m , ssa. Three specimens in the collection of the Institute of Vertebrate Biology of the Czech Academy of Sciences (collected in 2012 by J. Bryja, R. Šumbera, and Y. Meheretu) : ETH 0001 and ETH 0002 , both from the Menagesha Forest , edge of Juniper Forest, 2878 m a.s.l. ; ETH0234 , Sodo Hotel Garden , 2062 m a.s.l. ZMMU S-167296 (adult male, skull and dry skin, collector’s number LAV0944 ) was collected by L.A. Lavrenchenko in the Sheko Forest (07° 04′ 00″ N, 35° 30′ 00″ E, 1930 m a.s.l.) on 22 March 1999 .

Description: The dorsal pelage carmel brown with basal 80% slate grey, a faint stripe typically visible on dorsum, the venter including chin and throat snow white to roots, white sub-auricular ear tuft typically absent and modest when present, no dorsal head spot, ears blackish with orange and black hairs, tail modestly bicolor and much longer than head and body (120–140%), dorsal hairs of feet white but some with peripheral hairs more orange, claw on pedal digit V (as in Dendromus sensu stricto), hind foot long (18–19 with claw), well defined manual digit V, teats 4 + 4; Incisive foramen penetrates ca 40% length of 1st molar, incisors opisthodont, no sub-squamosal foramen. Palatal ridges: 3 complete (pre-dental), 4 divided (inter-dental).

Genetically confirmed specimens from the southern part of its range (Menagesha Forest, Sheko Forest, and Sodo) are larger and genetically distinct from the topotypical series from the northern part of their range (Simien Mts). Specimens from the south are in average 18% larger in HB (77.5 vs 65.37) and body mass (11.375 vs 9.6), with proportionately shorter tails (123% longer than HB vs. 142%). They are 10% larger in certain cranial dimensions (GLS 22.8 vs 20.8), with longer nasals (LN 8.3 vs. 7.6) and a longer posterior palate (PPL 7.7 vs 7.1). Further investigation and specimen acquisition will determine if we recognize the southern specimens as a species new to science.

Diagnosis: The new species has a faint, but visible, dark dorsal stripe. The tail is bicolored. Ears are blackish with both orange and blackish hairs, but the black hairs can be absent. Hindfoot 18–19 mm with claw. Size is larger than true D. mystacalis , including the GLS (20.3–25.0) and nasals (7.3–9.1).

Comparisons: A member of the D. mystacalis group as defined by Voelker et al. (2021) and incorrectly listed therein as Dendromus mystacalis . Three of the seven forms listed for this group are cryptic and were likely co-mingled in the past ( D. mystacalis , D. sp. indet. 2, and D. sp. indet. 3). As members of Dendromus (sensu stricto), they all have a claw, not a nail, on their 5th find toe, and all have light brown dorsal pelage sharply contrasting the belly fur that is white to the roots from the chin down through the belly. The dark dorsal stripe visible if compared to true D. mystacalis , where it is typically absent. Tail bicolored compared to mostly unicolor in D. mystacalis .

This new species is distinct from the other six members of the D. mystacalis clade (sensu Voelker et. al., 2021). Dendromus lovati , an Ethiopian endemic, is the only member of Dendromus in Africa described with four dorsal stripes, whereas this new form has a very faint single stripe. Both D. lovati and D. abyssinicus have belly fur with grey roots, unlike the other 5 members of this clade, which have belly fur white to the roots. Both D. ruppi and D. sp. indet. 4 (described below) have prominent dorsal stripes. Among the species with which it might be confused are true D. mystacalis , discussed above, and D. sp. indet. 2 ( Voelker et al., 2021). Since we are unable to describe D. sp. indet. 2 due to inadequate material, we compare the new species to D. mystacalis .

From D. mystacalis , this species is distinguished by its large body size, and long hind feet with claw (17.6–19.6), compared to 16–17 in D. mystacalis ). The new species has larger craniometrics than true D. mystacalis . GLS 20.3–25.0 (vs. 19.8–20.75), nasal 7.3–9.1 vs 6.4–7.3, M1-M1 breadth 4.57–5.0 (vs. 4.36–4.52). Incisive foramina reach the T4 of M1 whereas in D. mystacalis they fall just short.

Taxonomic notes: Dieterlen (2009) revised the Ethiopian material and assigned most records of the white-bellied Dendromus to D. mystacalis . We sequenced the holotype of D. mystacalis and showed that it represents D. sp. indet. 3, ( Voelker et al., 2021) and probably constitutes only part of the specimens analysed by Dieterlen (2009). Dendromus pseudomystacalis , sp. nov., was reported as mystacalis in the largest phylogenetic study of the genus ( Voelker et al., 2021), and a majority of the records of specimens with dark dorsal stripes and white bellies reported by Dieterlen (2009) as D. mystacalis from Ethiopia also likely belong to this species.

The taxonomic relationships between D. pseudomystacalis sp. nov. and D. sp. indet. 2 (sensu Voelker et al., 2021) are not yet fully resolved. Here we keep them as separate taxa (divergent gene pools) restricted to the Ethiopian highlands (see below). These two taxa do not have sister relationships in mtDNA ( Fig. 2 ) nor in ddRAD phylogeny based on concatenated data ( Fig. 3 ); however, genomically, they are very similar and overlapping (see PCA in Fig. 4 ), and there are no obvious morphological differences between them. Whether there is a reproductive barrier (suggested by a multispecies coalescent analysis in BPP, see Results) between them requires further integrative studies based on additional samples.

Ecological notes: The Ethiopian D. pseudomystacalis , sp. nov., has a strictly montane distribution ( Fig. 9 ). According to Yalden et al. (1976), it [as D. mystacalis ] occurs in the vegetation zone of Woina Dega, mid-elevation (ca. 1500–2300 m) highland, and the lower montane zone but is always associated with long grass and bushes. In Simien Mountain National Park, it has been reported from giant health ( Erica arborea ) forests with thick undergrowth ( Muller, 1977; Muhmenthaler, 1999; Wehrli, 1999; Craig et al., 2020). The Mekelle University/Field Museum expedition to the Simien Mts of Ethiopia (2015) surveyed habitats between 2900 and 4000 m a.s.l. ( Craig et al., 2020). Ten D. pseudomystacalis , sp. nov., were captured exclusively from Sankaber camp (3250 m a.s.l.). A second species of Dendromus in the Simien Mts., D. lovati , was found in all four camps (2900, 3250, 3600, 4000 m a.s.l.). In the Simien Mt. NP, the new species was only caught in the transition zone between the Afromontane forest and the ericaceous heath with Hypericum revolutum ( Craig et al., 2020) . At Menagesha forest, it was captured at the edge of Juniperus forest. In Sodo, it was collected from a hotel garden amid grass and trees. In the Sheko forest, the new species was captured in a disturbed humid Afromontane forest with a notable abundance of parasitic Ficus and undergrowth dominated by Coffea arabica .

Distribution: According to Dieterlen (2009), the distributional limit spans a broad range of historical occurrences. This agrees with our genotyping, making it the most widely distributed species of Dendromus in Ethiopia ( Fig. 9d ). The collecting localities are scattered over a vast region, covering large parts of the western and central highlands ( Bryja et al., 2019; Yalden et al., 1976). Our genetically confirmed records (only from the Abyssinian Plateau) span the elevation from 1930 (Sheko forest in southwestern Ethiopia) to 3259 m a.s.l. (Sankaber in Semien Mts.). This is a very common pattern for other widespread species in the Ethiopian highlands. They usually reach their lower elevational limits in southwestern Ethiopia, and their highest elevation records are from the much drier mountains of northern Ethiopia (see, e.g. Stenocephalemys albipes or Desmomys harringtoni ; Bryja et al., 2019).

Etymology: The specific epithet is derived from D. mystacalis and the Greek prefix pseudo -, meaning “false D. mystacalis ”. The species name reflects our long confusion of this species with “true” D. mystacalis s. str. and a substantial similarity between these taxa.

5) Dendromus sp. indet. 2 (sensu Voelker et al., 2021)

Dendromus mystacalis (part.) ( Yalden et al., 1976: distribution)

Dendromus mystacalis (part.) ( Dieterlen, 2009: distribution)

Dendromus sp. indet. 2 ( Bryja et al., 2019: distribution)

Dendromus sp. indet. 2 ( Voelker et al., 2021: phylogeny)

Taxonomic notes: We have five genotyped specimens belonging to this genetic clade: two from the Mekelle University campus (MEKxAb60, MEKxAb72), two from the Chercher Mts. in the eastern part of the Harar plateau ( ETH0080 , ETH0103 ), and one from Mekane Selam, close to the Borena Saynt NP ( ETH1340 ). Unfortunately, we had only the two specimens from the Chercher Mts. available for morphological examination. The remaining three specimens were lost in Mekelle during the recent civil war (2020–2023). This taxon may be conspecific with D. pseudomystacalis sp. nov., (see PCA in Fig. 4 ), but we keep them provisionally as separate taxa. This opinion that they are separate species is supported by the results of multispecies coalescent analysis in BPP (two confirmed divergent gene pools), and they also do not have sister relationships in mtDNA ( Fig. 2 ) nor in ddRAD phylogeny based on concatenated data ( Fig. 3 ). On the first view, there are no obvious morphological differences between them, but our material is very limited .

Distribution and ecological notes: The two individuals from the Mekelle University Campus (2200 m a.s.l.) were collected in grassland at the nursery site ( Welegerima et al., 2024). The specimen from Mekane Selam (2896 m a.s.l.) was collected from tall Eucalyptus , while those from Asbe Teferi (2331–2427 m a.s.l.) were collected along degraded field margins on a slope. At all three localities, it was the only species of Dendromus present.

6) Dendromus ruppi Dieterlen (2009)

Dendromus ruppi Dieterlen, 2009

Dendromus mesomelas subsp. ( Setzer, 1956)

Holotype: SMNS27572 , skin (mounted on cardboard) and skull of an adult female; field number 447; collected on April 23, 1978, by Hans Rupp.

Type locality: Gilo, Imatong Mts., East Equatoria, South Sudan; altitude ca. 1800–1900 m a.s.l.

Taxonomic notes: Clearly larger and with a much longer tail than Ethiopian D. pseudomystacalis sp. nov. and other white-bellied taxa from the Ethiopian clade (except D. rogersi sp. nov., Kilimanjaro). Blackish mid-dorsal stripe is always present ( Fig. 7 ). Based on phylogenomic results ( Fig. 3 ), it is sister to Dendromus . sp. indet. 4. from Mt. Kilimanjaro. They together represent the “out-of-Ethiopia” lineage in the D. mystacalis clade of Dendromus .

Distribution: Endemic to Imatong Mts. in South Sudan ( Fig. 9d ). The species was described from an elevation above 1800 m a.s.l. A recent (2015) expedition to the Imatong Mts. provided four additional specimens, deposited at FMNH from Gabigabi, near Kinyeti River (2510 m a.s.l.); three of them were captured in pitfalls.

Ecological notes: The specimens in SMNS were collected in a swamp or so-called “potato farm” ( Dieterlen, 2009); the two MCZ specimens came from a “nest in an olive tree” ( Setzer, 1956). The species prefers higher elevations, where it might replace D. mystacalis (see above). The latter is a rather lowland species in South Sudan ( Dieterlen, 2009; as D. mystacalis lineatus ), but it also occurs on the foothills of the Imatong Mts.

7) Dendromus rogersi Kerbis Peterhans and Voelker , sp. nov.

Dendromus mesomelas ( Mulungu et al., 2008)

Dendromus melanotis ( Stanley et al., 2014)

Dendromus sp. indet. 4 ( Voelker et al., 2021: phylogeny)

Holotype: Field Museum of Natural History , Division of Mammals, number FMNH174192 (field number WTS 5203 ), collected by W. T. Stanley, 28 July 2002. The type specimen, consisting of the skin, skull, and skeleton, is an adult male with a fused basi-occipital suture and testes measuring 6 × 4 mm. External measurements were made in the field: tail length 111 mm, head and body length 82 mm, hindfoot length 20 mm, ear length 17 mm, weight 11 g. The specimen was caught in a pitfall.

Type locality: Tanzania, Kilimanjaro Region, Moshi District, Mt. Kilimanjaro National Park , 10.5 km N and 3.5 km W of Maua (3° 10.627 S, 37° 26.413 E) at an elevation of 9500′ (2897 m a.s.l.) .

Paratypes: Nine additional specimens were captured on Mt. Kilimanjaro between July 19, 2002, and August 29, 2002 . FMNH174210 , f ; FMNH174211 , m ; FMNH174213 , m ; FMNH174214 , f ; FMNH174216 , m, all 5 at 2043 m a.s.l. ; FMNH 174188 , f, 2470 m a.s.l. ; FMNH174190 , m ; FMNH174194 , m ; FMNH174195 , f, all 3 at 2897 m a.s.l.

Diagnosis: Largest member of the D. mystacalis clade of Dendromus, sensu Voelker et al. (2021) . ONL mean of 24 mm (range 23.1–24.8), braincase breadth 10.7 mm (10.5–11.1), nasals elongate 8.9 mm (8.4–9.2). Incisive foramina shallow, not penetrating to T 4 of M 1. Ventral pelage, including chin and throat, snow white to roots. The tail is unicoloured and absolutely longer than other members of the clade (mean 114, 111–120). Ears are very long (17 mm), with a blackish peripheral ring. Slight sub-auricular white spot.

Comparisons: The new species exceeds all East African and Ethiopian members of the D. mystacalis group as defined herein mean body length, tail length, ear length, greatest length of skull, nasal length, and breadth of braincase. Our new species differs in its longer tail and ears and in its absolutely snow-white belly fur to the roots, whereas the Imatong form can be washed in orange buff. Dendromus rogersi sp. nov. has a longer skull than D. ruppi ( ONL mean 24 vs 22.8 mm), mostly due to nasal bones that are ca. 10% longer in the former (8.9 mm vs. 8.1 mm). In addition to its larger size, four cranial characters can distinguish this from D. ruppi : (1) the upper half of the zygomatic plate is more vertical (orthogonal) vs. sloping to the rear; (2) the incisive foramina fall short of the lingual bulge of the M1( T 4) vs reaching the T 4; (3) the maxillo-palatal suture falls between the M1 and M2 vs. at the beginning of the M2; (4) there is a tiny but visible sub-squamosal foramen in the new form, vs. no visible foramen. The braincase is broader and more fully inflated than in D. ruppi . The Kilimanjaro form is a forest denizen, while the Imatong form inhabits bushland and grassy habitats as well as agricultural plots.

Description: The new species is a member of Dendromus sensu stricto (as opposed to Poemys spp. ), as it displays a long claw on pedal digit V. The black dorsal stripe is well-defined from between the shoulders down to the base of the tail. Dorsal head spot absent. Small white spots (sub-auricular tufts) beneath the ears. Dorsal pelage is clear caramel and not mottled; ¾ of dorsal pelage with blue-grey slate, distal 25% a clear caramel brown. Vibrissae black, 20–28 mm. Small black patch on the lateral aspect of the foot down to the ankles. Hairs of foot caramel brown. The tail is very long (140% of head and body), and more or less unicoloured. Ventral fur snow white to roots, including chin and throat. Ears very long (17 mm), clothed in dark brown hairs within and without and with blackish fleshy rim. The thumb consists of a well-defined stump. Very large ( HB 76–86 mm) with a long tail (avg 114; 110–120 mm). The skull is very large ( GLS ca. 24, CI ca. 21.3, CLM ca. 3.5). Incisive foramina penetrates ca. 20% down the first molar, typically falling short of T 4. Post-palatal foramina lie within the front 1/2 of M2. Palatal suture typically falls between M1 and M2. Molariform teeth are selenodont rather than bunodont.

Distribution: At present, it is known only from Mt. Kilimanjaro, in a forested zone between 2043 and 2897 m a.s.l.

Ecology: All specimens were captured in the lower and upper forest zones of the Mt Kilimanjaro, where they are sympatric with D. insignis kilimandjari . The Field Museum expedition (Stanley & Rogers) to Kilimanjaro in 2002 surveyed habitats between 2000 and 4000 m a.s.l. Forty-six Dendromus spp. were captured: D. insignis kilimandjari was the most common (n = 36); they were captured at each camp (2000 m, 2500 m, 3000 m, 3500 m, and 4000 m). Dendromus rogersi sp. nov. (n = 10) was only captured in the lower and upper montane forest zones at 2000 m, 2500 m, and 3000 m ( Stanley et al., 2014). The two species were syntopic between 2000 and 3000 m a.s.l.

Reproduction: Of the 4 females captured (season of capture, July 19–August 29, 2002), 3 were imperforate; the 4th was pregnant (1 + 2) with embryos, crown-rump length 8 mm. The condition of the mammae is not visible. Of the six males captured, two had scrotal testes (9 × 5 mm) and three with abdominal testes (1.2 × 3 mm, 2.8 × 5 mm).

Discussion: It remains a puzzle how this mouse could have remained undiscovered and/or unrecognized on one of the planet’s most iconic and well-studied mountains. Dendromys nigrifrons = Poemys nigrifrons ( True, 1892) was the first taxon described from the Kilimanjaro area, represented by five specimens collected at 5000′ (1500 m a.s.l.) from the arid lowlands. Four decades later, a second taxon was described from 3800 m a.s.l., Dendromus mesomelas kiimandjari ( Bohmann, 1939) . Phylogenetically, D. m. kilimandjari nests comfortably within the D. insignis group ( Voelker et al., 2021, Fig. 1 : FMNH samples FMNH 174189, FMNH 174197, FMNH 174198, FMNH 174199, FMNH 174206).

There have been several reviews of the mammal fauna of Kilimanjaro ( Grimshaw et al., 1995; Swynnerton & Hayman, 1951). These authors pointed out the elevational distribution of two forms ( P. nigrifrons and D. kilimandjari ), whereas a compendium by Child (1965) simply listed the two generic “species” without reference to a definitive taxon (i.e. D. melanotis , D. mesomelas ), elevation or habitat. Several expeditions simply did not catch any Dendromus (e.g. the C.D. Rudd expedition; Thomas, 1910); and others did not sample an adequate range of habitats and elevations. For example, Shore and Garbett (1991) restricted their collecting to 3500 m a.s.l. where they only collected D. insignis kilimandjari . In any case, the literature seems to indicate that there were two forms: one, D. melanotis ( D. nigrifrons ) at lower elevations and another, D. mesomelas ( D. kilimandjari ) at higher elevations. In modern times, both Mulungu et al. (2008) and Stanley et al. (2014) conducted thorough transects (880–3600 m a.s.l. and 2000–4000 m a.s.l., respectively) and both collected two species of dendromurines. Both authors referred specimens from higher elevations (ericaceous zone and alpine zone, 2720–4000 m) to D. insignis . Mulungu et al. (2008) erroneously referred lower elevation specimens from 2350–2600 m a.s.l. to D. mesomelas , which does not have belly hairs that are pure white to the roots. Likewise, Stanley et al. (2014) incorrectly referred lower elevation specimens from the montane forest, (2040–2900 m a.s.l.) to D. melanotis ; however, Poemys melanotis has black ears, a nail on pedal V and a black head spot ( D. nigrifrons ), none of which are shared with the specimens described here. We would guess that Mulungu et al.’s lower elevation forms, housed at the University of Antwerp, would belong to the species described here. To sum up, there are three forms of dendromurines associated with Mt Kilimanjaro: one in the lowland savannah and scrub below the forest zone ( Poemys nigrifrons , 1500 m a.s.l.), a second ranging from the forest up through the ericaceous zone (2040–4000 m a.s.l., D. insignis kilimandjari ), and the new form we describe here ( D. rogersi sp. nov.) in mid-elevation between the former two and restricted to the forest zone (2040–2900 m a.s.l.).

Etymology: We named this form in honour of Mary Anne Rogers, a long-serving staff member and associate of the Field Museum of Natural History. She collected part of the hypodigm of this new species on Mt. Kilimanjaro. An accomplished mammalogist, she accompanied the late WT Stanley on many of his African expeditions, including this one. Her meticulous note-taking, attention to detail, and passion for fieldwork have been instrumental in bringing these critical collections to light.

Figures

Fig. 9 The distribution of the D. mystacalis clade across the known geographic ranges (the genotyped material corresponds with the colour bar on the phylogenetic trees in Figs. 2 and 3). Type localities are marked by asterisks. Non-genotyped specimens are represented by white circles. D. lovati; non-genotyped individuals were assigned to this species on the basis of typical fur coloration (see more details in Meheretu et al., 2022) (a); D. sp. indet. 1; non-genotyped specimens are reported by Dieterlen (2009) as D. insignis; resurrected to D. abyssinicus from synonymy here (b); D. sp. indet. 3 (sensu Voelker et al., 2021); assigned as true D. mystacalis here, the white circles represent non-genotyped material referred by Dieterlen (2009) as D.

Fig. 8 Variation in the four external body measurements displayed by violin plots: weight (a), lengths of tail (b), ear (c), head and body (d), hind foot (e), and relative tail length (f). The central white dot denotes

Fig. 2 A Bayesian phylogenetic tree of the D. mystacalis clade (rooted by Megadendromus nikolausi) calculated from the mitochondrial CYTB sequences. The black dots indicate posterior probability

Fig.6 Species tree estimated in BPP under the multispecies coalescent model. The branches are represented by rectangles with lengths equal to differences between their bounding divergence times (Δ) and widths equal to population sizes divided by two (∕2) to make them directly comparable with divergence times

Fig. 3 A maximum-likelihood (ML) tree of Dendromus inferred from the 60% occupancy dataset containing 9359 loci; 9329 of them (99.7%) contained at least a single biallelic parsimony/informative

Fig.4 The principal component analysis (PCA) of genomic variation of the D. mystacalis clade conducted on binary-coded biallelic SNPs. Dif- ferent colours correspond to seven main lineages, as shown in Figs. 2 and 3

Fig.7 Dorsal (above) and ventral (below) views of the coat coloration of the D. mystacalis clade (sensu Voelker et al., 2021). From left to right: D. lovati (LAV1923), D. sp. indet. 1 (LAV2148) resurrected to D. abyssinicus from synonymy here, D. sp. indet. 3 (LAV2548), assigned as true D. mystacalis here, D. sp. indet. 2 (ETH0080)

Fig.1 Distribution of the D. mystacalis clade. Yellow circles indicate the localities with genetically characterized individuals; names in blue colour represent major mountain ranges and the Turkana basin. Dashed black lines show the approximate borders of the Great Rift Valley (GRV), separating the Abyssinian and Harar plateaus of the