The STATIS compromise (Supplementary Fig. 1) reveals that Athene noctua has a more prominent load than others in structuring the results (correlation circle with data sets), because of the particular lateral, narrow extensions of the supra-orbital processes in this species, absent in the 10 other crania used in this study. The PC1s of dorsal and lateral analyses group together, as do their PC2s (correlation circle with PCA axes), showing that these two analyses summarize the same information (see Supplementary Figs. 2 and 3).
Among PCs, LAT1 summarizes most of the deformation across all species except A. noctua, whereas DORS1 summarizes better (as a single PC) deformation between O. sunia and its descendent O. murivorus (Supplementary Fig. 4). Deformation from O. sunia to O. murivorus is also visible (shared between PC1 and PC2) in lateral view (Supplementary Fig. 4). In lateral view analysis, PC1 explains 33.5%, and PC2 22.7%, of the total variance. In dorsal view analysis, PC1 explains 53.4%, and PC2 27.1%, of the total variance. In both dorsal and lateral analyses, the broken stick model indicates that PC1 and PC2 are sufficiently informative.
The main deformation across the species (except A. noctua and its supra-orbital processes extensions) is correlated with body size, with small owls near one end, and large owls (Bubo) near the opposite end, along LAT1. This deformation, from large to small owls, primarily concerns the following: braincase expands proportionately in length relative to orbits, which become more frontal; the basicranium shifts rostrally; and the frontal region changes from flat and extended rostrally, to round and shifted toward the basicranium, giving a round forehead to small owls (Supplementary Figs. 2, 3). Deformation between O. sunia and O. murivorus, on the other hand, partly differs and as visible along DORS1 it consists of slight narrowing of braincase and widening of frontal region (interorbital). As visible in lateral aspect, deformation between O. sunia and O. murivorus consists of relatively slight lowering of the dorsal curve of the braincase, more lateral shifting of the caudal limit of orbit, and caudal shifting of the basicranium and ventral edge of interorbital septum.
Analysis of traditional measurements
The PCA analysis of all measurement data (except angles), transformed using GMs, yields a rather equilibrated distribution of specimens and variables across the four quadrants. The broken-stick model showed that only PC1 and PC2 bore significant information (Supplementary Fig. 5). PC1 accounts for 47.8%, and PC2 27.9%, of the total variance (hence altogether 75.7%) (PCA statistics are in Supplementary Tables 1–3). Otus murivorus groups with B. cinerascens and B. zeylonensis in the same quadrant (Supplementary Fig. 6A).
The specimens scores along PC1 show no correlation to their geometric means (Supplementary Fig. 7, r2 = 0.025). This implies that the part of variance expressed by PC1 is non-allometric in general, which includes the fossil. The results will therefore be interpretable in terms of causes other than allometric scaling, i.e., adaptation, and/or correlated evolution of adaptive with non-adaptive trait(s) 19.
The correlation circle (Supplementary Fig. 6B) shows the contribution of the variables to the morphological variation and so the factors affecting O. murivorus location in the morphospace. The loadings of variables on PC1 and PC2 (Supplementary Fig. 6B, Supplementary Table 3) help make a hierarchy among the factors affecting O. murivorus. Compared to O. sunia, O. murivorus is rather distant (Supplementary Fig. 6A), and exhibits relative cranium thickening, longer olfactory bulb (OB), and wider frontal (interorbital) region (and larger body size), among the more prominent variables. Furthermore, O. murivorus exhibits a relative reduction of the wulst (W), as well as differences in global brain (Br) dimensions (brain volume, surface, length, width), foramen magnum size, and the lengths of semi-circular canals (SCs). A few variables show too small loadings on PC1 and PC2 (< 0.1) to be interpretable from PCA alone (Supplementary Fig. 6B, Supplementary Table 3). They are further assessed using other analyses so as to determine which ones are really insignificant taken individually. The more prominent variables (loadings > 0.1) are all considered in detail.
Based on measurements and observations (Supplementary Table 4), bivariate or trivariate analyses and boxplot analyses help characterize O. murivorus (see below). Within each category (cranial parts; endocranial parts; measures of angles), the more contrasted features in O. murivorus, as they appear in the PCA, are here listed first.
Cranium relative size
The cranium of O. murivorus appears smaller than in all the other strigids studied here, relative to body size (Fig. 2, Supplementary Fig. 8). O. murivorus stands significantly and well apart from the allometric trend, being the species with the relatively smallest cranium of the dataset. O. murivorus also exhibits cranium size reduction compared with O. sunia (Fig. 2, Supplementary Fig. 8). There is an allometric trend of proportionately smaller crania in larger owls (cf. slope in Supplementary Fig. 8), but the noticeable decrease in relative cranium size from O. sunia to O. murivorus largely exceeds that expected in view of the body size increase of the latter.
Relative thickness of cranium bone wall
Otus murivorus exhibits a high relative thickness of braincase (Figs. 2, 3B–D). Thicknesses of cranium wall measured at higher and lower level on the sides (wulst and cerebral hemispheres region) yield similarly much higher ratios in O. murivorus than in other owls including O. sunia. In contrast, some extant species exhibit noticeably thin walls, i.e., O. scops, B. scandiacus, and B. bubo (wulst region only, not cerebral hemisphere region, for the latter species) (Supplementary Table 4).
Width of frontal region
Interorbital width, as also appears on PCA results, is higher in O. murivorus than in other owls (relative to cranium dimensions); next is B. zeylonensis (Supplementary Table 4).
Optic and trigeminal nerves foramina
There is no modification in the relative size of optic nerve foramen (visual sense) in O. murivorus compared with O. sunia and the other owls used in this study (Supplementary Table 4). The relative size of the maxillomandibular foramen for the trigeminal nerve V2-3 (tactile sense)20 is low in O. murivorus compared with the other strigids, and slightly lower than in O. sunia (Supplementary Table 4).
In O. murivorus, the endocranial volume, relative to body size, is markedly lower than in any other owl; this is visible on a scatter plot of LOG brain volume to LOG body mass with a diversity of strigid owls (Fig. 3A) and on boxplots (Fig. 2). Otus murivorus is significantly well below the values of the small-brained extant strigids, and attests to an important relative reduction compared with O. sunia. There is an allometric trend of proportionately smaller brains in larger owls (cf. slope in Fig. 3). The relative decrease in relative brain volume from O. sunia to O. murivorus largely exceeds that expected in view of the body size increase of the latter. In addition, the foramen magnum length in O. murivorus is relatively slightly reduced compared with O. sunia (cf. PCA, and Supplementary Table 4).
The relative development of the olfactory bulb in O. murivorus (Fig. 4B; here compared with O. sunia, Fig. 4C, and Athene cunicularia, Fig. 4D) is by far the highest of all strigids (Figs. 2, 4A), and stands among the highest ratio values of all birds22. It is comparable to those birds with the greatest olfactory sense (e.g., some procellariiform seabirds and some vultures) 22. Bubo cinerascens is intermediate between the average values of other owls and that of O. murivorus. Compared with O. sunia, O. murivorus also deviates significantly and strongly from the nearly isometric trend in Fig. 4A.
Other endocranial regions
Ratios of the surface of precise endocranial areas to the total endocast surface add evidence that O. murivorus deviates from other strigids in several features (Figs. 2, 5). Proportionately, O. murivorus significantly bears a greatly reduced wulst area, as well as a slightly enlarged cerebellum (Crb) compared with all other strigids examined, including O. sunia.
Conversely, O. murivorus exhibits an optic lobe (OL) surface ratio similar to that of other strigid owls (Fig. 2), and there is no apparent correlation between the surface of the optic lobe and nocturnal or diurnal habits in extant owls (with the more diurnal owls being in the sample B. scandiacus, B. zeylonensis, A. cunicularia and A. noctua23). Similarly, the cerebral hemisphere (CH) ratio is not significantly different in O. murivorus compared with O. sunia and other strigids (Fig. 2).
Considering the cochlear duct, a ratio of its length (see Supplementary Table 4) to the cranium height ranges from 0.21 (B. scandiacus, B. cinerascens) to 0.31 (M. asio). The difference is slight between O. murivorus (0.25) and O. sunia (0.28). The lengths of semi-circular canals show more contrasts between species. On a 3D plot, where each axis measures a ratio of one semi-circular canal length to cranium height, the two long-distance migrant scops-owls O. sunia and O. scops group together with high ratios. A central group contains O. murivorus and all remaining species, except B. zeylonensis, which exhibits the lowest ratios (Supplementary Fig. 9). O. murivorus exhibits a decrease in all semi-circular canals relative lengths compared with O. sunia (Fig. 2, Supplementary Fig. 9). The relative thickness of semi-circular canals16 varies quite widely among strigids, and it is medium in O. murivorus, but proportionately higher than in O. sunia (see Supplementary Table 4, Fig. 6B–E). The sinuosity of the lateral semi-circular canal (LSC) 16 varies as well. In O. murivorus, it is rather medium, hence greater than in O. sunia, which exhibits the flattest LSC along with O. scops (the two long-distance migrating owls of the sample) (Fig. 6B–E; Supplementary Table 4).
Measured angles (inner ear and cranium)
When plotted in 3D, the three values of angles between semi-circular canals two by two yield three groupings (Fig. 6A). Otus murivorus groups with A. cunicularia, with medium PSC/LSC angles, high LSC/ASC angles, and the lowest ASC/PSC angles. A group contrasting essentially with moderate to high ASC/PSC angles contains all other owls except B. scandiacus. The latter stands at the opposite of A. cunicularia and O. murivorus, with high ASC/PSC angle, and the lowest PSC/LSC and LSC/ASC angles (Figs. 2, 6A).
Orientation of the orbits
In terms of orbital margin convergence, O. murivorus exhibits an angle (37°) that attests to a strong decrease compared with O. sunia (44.9°). In other words, the orbits are placed more laterally. However, among the owls of the extended sample using data from Menegaz et al. 26, O. murivorus is in a rather intermediate position, compatible with both nocturnal and diurnal owls, given the wide dispersion of both (Supplementary Fig. 10), and despite there being a slight tendency for lower angles in diurnal species. Otus murivorus exhibits an angle more close to the nocturnal mean than to the diurnal mean, but the discrepancy between the two groups is low.
In terms of posture, as measured as a relative position of the foramen magnum on the cranium, the higher angle in O. murivorus compared with O. sunia amongst others reflects largely an allometric relation (Supplementary Fig. 11); this angle is higher in large owls (small owls have more ventral foramen magnum). This angle is still slightly high in O. murivorus, as it is in B. zeylonensis for example.