Results

A total of 56 skulls and 50 mandibles of R. simulator and 19 skulls and 14 mandibles of R. cf. simulator were analysed. Procrustes ANOVA tests did not find significant differences between sexes in both species (both in size and shape of skulls and mandibles). For R. simulator skull size, F 1; 54 = 2.19, p = 0.14; skull shape, F 65; 3510 = 1.01, p = 0.46; mandibles size, F 1; 49 = 4.77, p = 0.03; mandible shape F 38; 1862 = 2.47; p< 0.0001. For R. cf. simulator skull size,F 1; 17 = 1.25, p = 0.28; skull shape,F 65; 1105 = 0.93, p = 0.64; mandible size,F 1; 12 = 3.97,; p = 0.07, mandible shape,F 38; 456 = 0.79, p = 0.81. Sexes were therefore pooled for all analyses, balancing the number of males and females for R. simulator mandibles.
There was variation in the shape of skulls across different localities within each lineage (R. simulator : F 195; 3445 = 2.22; p < 0.001, R. cf. simulator :F 130; 1040 = 2.37; p < 0.001) but not in size (R. simulator skulls: F 3; 53 = 0.15, p = 0.93; R. cf. simulator skulls:F 2; 16 = 2.57; p = 0.11). The mandibles ofR. cf. simulator differed in shape across localities (F 76; 418 = 1.52; p < 0.01) but not size (R. cf. simulator mandibles: F 2; 11 = 1.68; p = 0.23). Those of R. simulator were not different in both shape (F 114; 1786 = 0.38;p = 0.15) and size (F 3; 47 = 0.17;p = 0.91).
Skulls
For R. simulator , the first two canonical variates of the canonical variate analysis (CVA) of shape variation amongst the localities of R. simulator explained a total of 90% of the variation (Fig. 1). The wireframe graphs (Fig. 1) show that the first canonical variate (CV1) was associated with changes in the palate, zygomatic arch, cranium and cochlea structure (66% of the variation). Skulls from the NZ locality fell at the positive end of CV1 and appeared to have a wider zygomatic arch, broader cochlea and longer palates relative to the average. Conversely, skulls from the SE locality fell at the negative end of CV1 and had a reduced zygomatic arch, a narrower cochlea and shorter palates relative to the average. Two localities (NE and SZ) fell within the intermediate zone of the CV prescribed shape space, implying that it had a shape close to the average. CV2 was mostly associated with the anterior medial swelling (24% of the variation; Fig. 1). The SZ locality fell at the positive end of CV2 and had an outline implying increased volume of the nasal dome relative to the average. Skulls from two localities (NE and NZ) fell at the negative end of CV2 indicating that they had a smaller anterior medial swellings than the average, and skulls from one locality (SE) were positioned intermediately along CV2 indicating that it had an anterior medial swelling close to the average. CV3 (S2 Fig) explained 10% of the variation and was associated with changes in the zygomatic arch and palate. Skulls from NZ and SE were on the negative end of CV3 suggesting that they had broader zygomatic arches, and longer palates relative to the average shape and the position of SZ and NE along CV3 indicated that these skulls had narrower zygomatic arches and shorter palates relative to the average shape (S2 Fig).