Myosorex tenuis

Taylor, Peter John, Kearney, Teresa Catherine, Kerbis Peterhans, Julian C., Baxter, Roderick M. & Willows-Munro, Sandi, 2013, Cryptic diversity in forest shrews of the genus Myosorex from southern Africa, with the description of a new species and comments on Myosorex tenuis, Zoological Journal of the Linnean Society 169 (4), pp. 881-902 : 898-899

MYOSOREX CF. TENUIS THOMAS & SCHWANN 1905 THIN FOREST SHREW OR TRANSVAAL

FOREST SHREW

Holotype

BM 4.9 .1.22. External and cranial measurements are shown in Table 3. Type locality is Zuurbron, near Wakkerstroom in Mpumalanga Province, South Africa (indicated in Fig. 1 ).

Referred material: See the Appendix. Diagnosis

This species is clearly differentiated genetically and biogeographically from all other southern and eastern African species of Myosorex ; however, it is difficult to diagnose morphologically as pelage, craniodental, and size characters vary considerably within the species. Nevertheless, from series examined in this study, the condition and position of the fourth unicispid (most particularly the pronounced extension of the parastyle of the upper anterior premolar, which results in a narrow gap between this tooth and the upper third unicuspid) is similar to that found in M. meesteri sp. nov., and clearly differentiates M. cf. tenuis from other South African species ( Fig. 6 ; Table 7). The consistency of this character should be tested from larger and geographically broader samples. Typically, specimens from north Drakensberg (Wakkerstroom, Woodbush) and west Soutpansberg are easily distinguished by their small cranial size, with CI usually 20–22 mm, as opposed to> 22 mm in M. varius and M. cafer ( Table 3). The Zuubron, Wakkerstroom type specimen is sympatric with the larger-sized M. varius ( Thomas & Schwann, 1905) ; however, populations from Woodbush and Soutpansberg occur allopatrically. Populations from east Soutpansberg are larger in size, and overlap in cranial and external measurements with M. varius ; however, these populations are genetically associated with M. cf. tenuis from western Soutpansberg rather than with M. varius ( Fig. 2 ), and they tend to have a darker hindfoot colour and unicoloured tail more typical of M. cafer (from which they are also distinguished genetically). Roberts (1951) indicated clearly that M. tenuis was a dark-footed form and included Wakkerstroom and Woodbush in its range; however, examination of a large series from Woodbush indicate that they are all relatively pale-footed ( Table 7). Wolhuter (in Smithers, 1983) pointed out that specimens from the type locality of M. tenuis had a karyotype (2 n = 40) that was distinct from those of M. cafer (2 n = 38) and M. varius (2 n = 42), and that was found in populations from ‘about Wakkerstroom’ to as far north as Entabeni in the central Soutpansberg. These data provide additional support for the existence of M. tenuis occurring from Wakkerstroom to Entabeni (Soutpansberg); however, the resolution of the correct name for this lineage awaits detailed molecular and morphological analysis of the holotype of M. tenuis in the London Natural History Museum, in comparison with critical historical and recent collections.

Description

Pelage colour varies considerably, from individuals (e.g. from Entabeni) that are dark (almost blackish) in dorsal colour, and with dark feet, like M. cafer , to others with brownish dorsal pelage and paler hindfeet, more like M. varius ( Table 7). Likewise, body and particularly cranial size varies dramatically with populations, falling into two divergent size classes (with minimal overlap between them): smaller-sized individuals from western Soutpansberg and the northern Drakensberg Escarpment, and larger-sized individuals from eastern Soutpansberg ( Figs 3B , 4 , 5 ). Palatal foramina are mostly overlapping (as in M. varius ), but are sometimes (8%) nonoverlapping or with the medial foramina absent (6%; Table 7).

Biology and distribution

Based on the material examined in this study, M. cf. tenuis is mostly restricted to the Limpopo Province of South Africa from the Soutpansberg Range to the northern extension of the Great Escarpment of South Africa, extending southwards to the type locality of M. tenuis (Zuurbron, Wakkerstroom District), which is located some 400 km south of Woodbush on the border between Mpumalanga and KwaZulu-Natal provinces ( Figs 1 , 5 ). It is strange that few or no museum specimens in the intervening northern Drakensberg region of Mpumalanga Province have been assigned unequivocally to M. tenuis , with collections from this region from the TM being referred mostly to M. varius (or apparently in error to M. cafer ). An accurate understanding of the distribution limits of M. cf. tenuis awaits a critical analysis of historical and recent collections (in the TM and NHM) of Myosorex from the Mpumalanga Drakensberg.

It appears that phenotypic and possibly genotypic differentiation is continuing in this species, as evidenced by the large morphological gap between populations west (Lajuma) and east (Buzzard Mount, Entabeni, Middelfontein, and Hanglip) of the Sand River Valley, which bisects the Soutpansberg from north to south. From recent Soutpansberg collections, these shrews were almost always collected in wetlands and moist grasslands, although a couple of individuals were collected from the margin of mistbelt forests. This habitat association further emphasizes the ecological separation between this species and the forest specialist M. cafer , with which it was until recently associated.

Figures

Figure 1. Map of southern Africa showing sampling localities for morphometric and molecular analyses of Myosorex. Grey shading represents the Great Escarpment of South Africa and the eastern Zimbabwean montane grassland–forest mosaic ecoregion of Olson et al. (2001). (Note: the Gorogosa locality overlies a small isolated patch of this ecoregion.) Symbols indicate recognized and newly defined species as follows: open and closed squares represent morphological and molecular sample localities, respectively, for Myosorex varius; open and closed triangles represent morphological and molecular samples, respectively, for Myosorex cafer; the hash symbols represent molecular samples of Myosorex sclateri; open and closed circles represent morphological and molecular samples, respectively, of Myosorex meesteri sp. nov.; crosses and asterisks represent morphological and molecular samples, respectively, of Myosorex cf. tenuis; ⊗, type locality (Zuurbron, Wakkerstroom District, Mpumalanga) of Myosorex tenuis. More details of the samples and localities are provided in Table 1 and the Appendix.

Figure 2. Maximum-likelihood phylogeny (left) and maximum clade probability tree (right), inferred from the combined analysis of molecular data (mitochondrial DNA control region, 16S rRNA, and the nuclear intron STAT). The newly resurrected or described species Myosorex tenuis and Myosorex meesteri sp. nov. are in bold. Maximum-likelihood bootstrap and posterior probability (in that order) values are shown for nodes of the maximum-likelihood tree. Values above nodes of the maximum clade probability tree indicate the posterior mean divergence dates in millions of years before present. Shaded bars indicate the 95% highest posterior density (HPD) credibility intervals. Values below the nodes indicate posterior probability values generated during the BEAST dating analysis.

Figure 3. Principal component analyses (PCA) of ten logtransformed craniometric variables in Myosorex samples from: (A) Zimbabwe (filled squares) and Mozambique (open squares); and (B) Limpopo Province, South Africa (■, Hanglip; ▲, Entabeni; △, Woodbush; ×, Lajuma; *, Buzzard Mount; ○, Middelfontein Farm). The first two principal components explained 40.8, 23.0, 62.5, and 16.1% of the total variance, respectively, for (A) and (B).

Figure 4. Canonical variate analysis (CVA) of ten logtransformed craniometric variables in six operational taxonomic units (OTUs) of Myosorex from southern and eastern Africa; ○, Zimbabwe–Mozambique; ●, Myosorex kihaulei and Myosorex geata (Tanzania); ▲, east Soutpansberg; △, north Drakensberg + west Soutpansberg; ×, Myosorex varius; +, Myosorex cafer (topotypic). The first two canonical vectors explained 72.7 and 14.7%, respectively, of the total variance.

Figure 5. Map showing distribution of two distinct morphological groups in the Soutpansberg Mountains and northern Drakensberg Mountains of Limpopo Province in relation to a map of annual precipitation (AP) for the region (pale-grey shading indicates AP of 800–1000 mm; dark-grey shading indicates AP of 1000–1300 mm; black indicates AP> 1300 mm). Open triangles represent the smaller-sized morph (from Woodbush and Lajuma), whereas open squares indicate the large-sized populations from east of the Sand River in the Soutpansberg (Buzzard Mount, Hanglip, Farm Middelfontein, and Entabeni Forest). The dashed line outlines the extent of the Soutpansberg Mountains.

Figure 6. Photographs of the fourth unicuspid and adjacent molars in the upper tooth rows of: (A) TM 10448, Myosorex sclateri (Ngoye Hills, KwaZulu-Natal); (B) TM 41824, Myosorex varius (Karkloof, KwaZulu-Natal); (C) TM 793, Myosorex cf. tenuis (Wakkerstroom, Mpumalanaga); (D) TM 25843, Myosorex cf. tenuis (Entabeni, Soutpansberg Range, Limpopo); (E) TM 34613, Myosorex meesteri sp. nov. (Mount Selinda, Chirinda Forest, Zimbabwe).