Malpaisomys, AND

Renaud, Sabrina & Michaux, Jacques, 2004, Parallel evolution in molar outline of murine rodents: the case of the extinct Malpaisomys insularis (Eastern Canary Islands), Zoological Journal of the Linnean Society 142 (4), pp. 555-572 : 560-562

COMPARISON OF MALPAISOMYS AND POSSIBLE

RELATIVES

Malpaisomys has been compared to different stages of the Progonomys – Occitanomys / Stephanomys and Paraethomys lineages, to Progonomys cathalai and to several modern taxa ( Table 1). An important morphological differentiation exists for the first upper and lower molars (P <0.001).

For the upper molars, the first two canonical axes display most of the among-group differentiation ( Fig. 3A ). In the plane defined by the first (CA1, 46.3% of the among-group variance) and the second axes (CA2, 18.2%), the first cluster includes Progonomys associated with Mastomys , Praomys and Malacomys . All of these murine teeth are characterized by an asymmetrical outline. They also share a basic omnivorous murine diet. From this ‘omnivorous’ cluster the second cluster diverges along both CA1 and 2, including murines with broader and more symmetrical molar outlines. Among these taxa, those with a known ecology share a trend towards a herbivorous diet. This cluster of ‘intermediate’ outlines includes Mio-Pliocene Paraethomys , Occitanomys , the arvicanthines (except Oenomys ), and Malpaisomys . From this cluster, Stephanomys (along CA1), the Pleistocene Paraethomys (mostly along CA2 with a component along CA1), and Oenomys (along CA2) show a further divergence. All these taxa are characterized by a typical stephanodont dental pattern associated with broader and more symmetrical outlines.

The modern taxa Praomys , Malacomys and to a lesser extent Mastomys ( Fig. 3B ) are especially isolated on the third axis (10.3% of among-group variance). It appears to correspond to a phylogenetic signal of the Praomys - Malacomys group.

The results based on the first lower molars ( Fig. 3C ) are less clear but suggest similar features than those based on the upper molars ( Fig. 3A, B ). The differentiation along CA1 (60.6% of the among-group variance) corresponds to a considerable divergence of Stephanomys , the most recent (LF4) being the most extreme. Along CA2 (20.5% of variance) Progonomys , associated with Malacomys , is opposed to Pleistocene Paraethomys . In between are found two clusters; the first includes the oldest Occitanomys and the Mio- Pliocene Paraethomys , Mastomys , Arvicanthis and Thallomys , the second includes more recent Occitanomys , Oenomys and Malpaisomys . CA2 therefore displays a trend from primitive towards more derived outlines, and Malpaisomys is associated with rather derived ones.

The distribution of the taxa on these canonical axes might appear to be a simplification of the morphological signal and thus undermine the interpretation of the morphological differentiation as an ecological signal showing evolutionary grades. We therefore complemented the multivariate analysis by a cluster analysis of the raw data, focusing on the first upper molar shape. Distances were calculated from the Fourier coefficients and a phenetic tree was constructed using a UPGMA algorithm ( Fig. 4 ). In agreement with the patterns observed on the canonical axes, outlines interpreted as ‘omnivorous’ are grouped together, including Progonomys , Malacomys , Praomys and Mastomys . Stephanomys appears as highly divergent. Derived, stephanodont Oenomys and Pleistocene Paraethomys are clustered and clearly separate from a cluster of outlines characterized by an intermediate stage of evolution: Occitanomys, Mio-Pliocene Paraethomys , the arvicanthines Aethomys and Arvicanthis , and Malpaisomys .

Figures

Figure 3. Morphological differentiation of the dental outline among murine rodents. A, upper molars. Group means on CA1 and 2. Dotted lines show the fossil lineages leading from Progonomys hispanicus to Occitanomys and to Stephanomys, and from Pr. cathalai to Paraethomys. Outlines correspond to some group means. B, upper molars, CA1 and 3. C, lower molars, CA1 and 2. For locality and sample abbreviations see Table 1.

Figure 4. Cluster analysis of the shape of the upper molar. Euclidean distances between group means were calculated based on Fourier coefficients up to the ninth harmonic (CF9). Clustering method was UPGMA (unweighted pair-group method, arithmetic average). Symbols as per Fig. 3. For locality and sample abbreviations see Table 1.