Hylomyscus, IN

Carleton, Michael D. & Stanley, William T., 2012, Species limits within the Praomys delectorum group (Rodentia: Muridae: Murinae) of East Africa: a morphometric reassessment and biogeographical implications, Zoological Journal of the Linnean Society 165 (2), pp. 420-469 : 450-451

AND HYLOMYSCUS IN TANZANIA

Populations of the Praomys delectorum group share their montane habitat with only one other member of Praomyini, Hylomyscus arcimontensis . Examples of H. arcimontensis co-occur with P. taitae at nearly all forested sites surveyed within the EAM ( H. arcimontensis not documented from the Taita Hills) and with P. melanotus in the Southern Highlands ( Stanley et al., 1998b [reported as H. denniae ]; Carleton & Stanley, 2005; Stanley & Goodman, 2011). Although distributions of the P. delectorum group and H. arcimontensis are broadly sympatric in Tanzania, the latter species has not been documented from the Northern Highlands ( Carleton & Stanley, 2005), nor has it been collected in moist forests of southern Malawi, the range of P. delectorum s.s. ( Hanney, 1965; Ansell & Dowsett, 1988; W. T. Stanley, unpubl. data). The patterns of craniodental differentiation evident among populations of Praomys and Hylomyscus are intriguing in light of the substantial congruence of their Tanzanian distributions, both of which span the Makambako Gap.

The following comparisons issue from DFAs of Praomys and Hylomyscus , conducted separately as conveyed in results reported above and those in Carleton & Stanley (2005) and based only on nine co-distributed montane samples (see Appendix 3 for specimens of H. arcimontensis ). The amount of variation captured by the first canonical variate extracted is greater for the nine-OTU DFA of Praomys (63.7%) compared with that derived for samples of Hylomyscus (46.2%). A pronounced shift in CV 1 scores occurs among Praomys samples, coincident with samples obtained on either side of the Makambako Gap; mean CV 1 scores substantially differ and ranges do not overlap between geographical subgroups. The nine samples of H. arcimontensis , on the other hand, broadly overlap in CV 1 ranges, and their means irregularly trend in a south-westerly direction from smaller (South Pare Mts) to larger (Mt Rungwe) animals; no abrupt shift in craniodental covariation is apparent between population samples on either side of the Makambako Gap ( Fig. 12 , top; also see Carleton & Stanley, 2005: fig. 8). For the nine montane samples of Praomys , the last clustering cycle based on Mahalanobis distances between group centroids deeply divides those in the Southern Highlands (Livingstone Mts, Mt Rungwe) from those in the EAM (South Pare Mts through the Udzungwa Mts); within the latter cluster, the OTU representing the Ukaguru Mts again emerges as the most strongly differentiated Praomys within the EAM ( Fig. 12 , bottom; also see Fig. 5 ). For examples of Hylomyscus from the same mountain systems, the last two clustering cycles successively amalgamate those samples from the northern periphery of the EAM (South Pare Mts, West Usambara Mts); the Hylomyscus OTUs from the Southern Highlands (Livingstone Mts, Mt Rungwe) do form a unique pair-group but at a much lower level of dissimilarity ( Fig. 12 , bottom). As argued herein, we recognize the pronounced craniodental divergence between Praomys populations separated by the Makambako Gap as demarcating separate species, P. melanotus (Southern Highlands) and P. taitae (EAM) . Carleton & Stanley (2005) interpreted the broadly overlapping samples of Hylomyscus as geographical variation within a single species, H. arcimontensis , including EAM localities as well as those from south-west of the Makambako Gap.

The morphometric contrasts between Tanzanian samples of Hylomyscus and Praomys surprised us. Having encountered only minor differentiation among Hylomyscus populations across the Makambako Gap ( Carleton & Stanley, 2005), we had anticipated a similar pattern within Praomys . Indeed, the distributional records for Hylomyscus intimate that its populations would be more likely to experience differentiation among montane isolates. Within the EAM, specimens of H. arcimontensis are unknown from elevations as low as those recorded for P. taitae ( Carleton & Stanley, 2005; Stanley & Hutterer, 2007, fig. 11): the lowest locality so far documented for H. arcimontensis is 900 m in the East Usambara Mts; whereas, examples of P. taitae are known from 230 m in the East Usambara Mts and 600 m in the Udzungwa Mts. Where collected in sympatry, populations of Praomys are more abundant, as measured by trapping success, than Hylomyscus and seem to better tolerate degraded secondary habitats (Stanley & Goodman, 2011). Compared with Praomys , H. arcimontensis is more capably scansorial and may be more dependent on arborescent primary habitats, as suggested by certain morphological adaptations (relatively long, penicillate tail and short, broad hindfoot with long fifth digit; Carleton et al., 2006), direct observations of climbing activity ( Stanley et al., 2005a), and comparative trap success in arboreal versus ground sets. Such considerations led us to view populations of Praomys as more prone to dispersion between isolated montane forests, facilitated by lowering of vegetation belts during Pleistocene fluctuations ( Hamilton, 1982; Lovett, 1993a), than those of Hylomyscus . The pronounced craniodental differentiation evident in the covariation of Praomys data, coincident with a major geographical barrier, contradicts this expectation.

Whether patterns of morphometric differentiation within Praomys vis à vis Hylomyscus mirror comparable levels of genetic divergence invites studies that employ a phylogeographical approach and incorporate several genes. Such a needed perspective will further illuminate differentiation among Praomys populations across these mountains and the interplay of geographical distance, isolation, and recency of diversification. We find it noteworthy that complementary molecular and morphometric data sets of soricomorph shrews distributed across these same Tanzanian mountains have disclosed different topologies of area relationship, one involving intraspecific variation ( Sylvisorex – Stanley & Olson, 2005) and the other addressing interspecific relationship ( Myosorex – Stanley & Esselstyn, 2010); in both generic examples, genetic similarity more closely tracked geographical proximity among EAM units than did the morphometric data set. The praomyine genera Hylomyscus and Praomys offer an ideal system to compare the diversification of rodents and shrews that cohabit Tanzania’s montane forests.

Figures

Figure 5. Cluster diagram (UPGMA) based on average Mahalanobis distances between the 21 OTU centroids as derived from discriminant function analysis (coefficient of cophenetic correlation = 0.938; also see Fig. 4). Bootstrap values are indicated for those nodes that were consistently defined (± 70%, 1000 iterations); mountain system and OTU identifier are indicated for each terminal OTU; species recognized herein are identified for those stems that subtend major geographical associations among the analytical samples.

Figure 12. Morphometric comparisons of samples of Hylomyscus and Praomys that co-occur in montane forest across the Eastern Arc Mountains (South Pare Mts to Udzungwa Mts) into the Southern Highlands (Livingstone Mts and Mt Rungwe) of Tanzania. Top graphs: north–south transects of morphometric differentiation as indexed by the first canonical variate extracted; vertical lines correspond to the sample range and filled circles to the sample mean (Abbreviations: EUS, East Usambara Mts; LIV, Livingstone Mts; NGR, Nguru Mts; RUN, Mt Rungwe; SPA, South Pare Mts; UDZ, Udzungwa Mts; UKA, Ukaguru Mts; ULU, Uluguru Mts; WUS, West Usambara Mts). Bottom phenograms: Cluster diagrams (UPGMA) based on Mahalanobis squared distances between the nine OTU centroids as derived from discriminant function analysis; only nodes that were consistently recovered by bootstrapping are indicated (± 70%, 1000 iterations), leaving poorly defined levels of phenetic affinity as unresolved. Both multivariate perspectives document abrupt shifts in craniodental differentiation between populations of Praomys coincident with the Makambako Gap (here interpreted as different species, P. melanotus and P. taitae) but not those of Hylomyscus (interpreted as the single species H. arcimontensis by Carleton & Stanley, 2005). See text for discussion.