A three-dimensional reconstruction of the skeleton of the giant rat of Tenerife (Canary Islands, Spain) Canariomys bravoi was obtained by computerized microtomography. Body size, body mass, and body shape were estimated, and limb morphofunctional indices used to infer the style of life of this recently extinct rat. A sample of recent Murinae, including the Philippines endemic giant cloud rat Phloeomys cumingi, was used for comparison. It appears that C.bravoi differed from most continental rats by its relatively large size, body proportions, and tail length. Among its distinctive features, claws almost similarly developed on fore and hind limbs, and feet longer than hands evoke an intermediate body shape between rats and arboreal murines like Phloeomys. C.bravoi was a strong and powerfully muscled rat able to move on different substrates from floor to trees, and probably had digging skills.
Une reconstitution en trois dimensions du squelette du rat géant de Tenerife (Îles Canaries, Espagne) Canariomys bravoi a été obtenue à partir d’acquisitions en microtomographie. Taille corporelle, masse corporelle, et forme du corps ont été estimées, et le style de vie fut déduit d’indices morphofonctionnels des membres. Des Murinae récents parmi lesquels Phloeomys cumingi, rat géant endémique des Philippines ont servi aux comparaisons. C.bravoi différait de la plupart des rats continentaux par sa grande taille, les proportions du corps, et la longueur de la queue. Parmi ses caractéristiques distinctives, les griffes presque d’égale importance aux membres antérieurs et postérieurs, et les pieds plus longs que les mains, évoquent une forme intermédiaire entre les rats et les vrais Murinae arboricoles comme Phloeomys. C.bravoi était un rat musclé, capable de se déplacer sur des substrats divers depuis le sol jusqu’au sommet des arbres, et capable de creuser.
Microtomography, Ecomorphology, Locomotion, Evolution, Insularity, Canary IslandsMicrotomographie, Écomorphologie, Locomotion, Évolution, Insularité, Îles CanariespresentedPresented by Philippe TaquetIntroduction
Mammals show high capacity for adaptation to environmental changes. They migrate or speciate, and their evolution may accelerate when colonizing islands. The Canary Islands, which are volcanic islands without any former or recent connection to the nearby continent, are of particular interest because they bear a good fossil record of recently extinct rodents. They form an archipelago close to the Tropic of Cancer, the easternmost islands of which are situated ca 100 km off the African coast at the level of the Sahara desert. This situation resulted in the settlement of an unbalanced fauna of terrestrial vertebrates, composed of a few genera of lizards and mammals, the latter being represented by three rodents and one shrew. All rodents were members of the Murinae (Old World rats and mice): C.bravoi on Tenerife (Crusafont-Pairo and Petter, 1964), Canariomys tamarani on Gran Canaria (López-Jurado and López-Martínez, 1991 and López-Martínez and López-Jurado, 1987), and Malpaisomys insularis on Lanzarote, Fuerteventura and some nearby islets (Hutterer et al., 1988). M.insularis was the sole rodent to be associated with another native mammal, Crocidura canariensis, a shrew still living today (Hutterer et al., 1987). The two Canariomys species were the size of a large rat while Malpaisomys was the size of a large mouse. All known fossils of the extinct rodents are Holocene or Late Pleistocene in age. These rodents, together with the giant lizards, became extinct during Prehispanic times as the result of human settlement (Atoche et al., 1995 and Navarro et al., 1990). The date of the native mammalian settlement on the Canary Islands is unknown due to the lack of fossils older than the Late Pleistocene (Michaux et al., 1996). Pliocene and Miocene sediments with terrestrial turtles (Hutterer et al., 1998) have yielded no mammals so far. The estimated molecular time of divergence of the endemic shrew provided scanty clues on the chronology of settlement of the endemic Canarian mammalian fauna. C.canariensis diverged from an African species around 1.61–4.30 Myr ago (Dubey et al., 2008). The present study still follows the interpretation of López-Martínez and López-Jurado (1987), who considered a Late Miocene divergence of the two species of Canariomys from the Arvicanthini (López-Martínez and López-Jurado, 1987 and Musser and Carleton, 2005). The molecular calibration of the Murinae recently provided a date of ca. 8.4 My for the separation of the tribe Arvicanthini, while the oldest fossils referred to members of the tribe are 6 to 7 My old (Lecompte et al., 2008).
C.bravoi has been known since the middle of the last century but has never been described in detail, except in concert with the description of C. tamarani given by López-Martínez and López-Jurado (1987). While these authors considered C. tamarani an herbivorous rodent with a terrestrial mode of life, including some digging skills, C.bravoi was interpreted as a generalist rat characterized by an omnivorous diet, quadrupedal locomotion, and likely to show some abilities for climbing trees. Later, several studies aimed at going deeper into the understanding of the palaeobiology of these extinct rodents. The massive shape of the mandible, which is very similar in the two Canariomys species, is better explained by the large size of these rodents rather than by a particular diet (Firmat et al., 2010). The stable isotope ratios indicated a diet dominated by vegetal matter in C. bravoi (Bocherens et al., 2003 and Bocherens et al., 2006). Microwear patterns also confirmed that the two species of Canariomys had a diet dominated by plant materials excluding grasses (Firmat et al., 2010). Surprisingly, whereas skeletal and tooth remains show differences that clearly support a specific distinction, C. bravoi and C. tamarani were likely to have had a very similar diet.
A better paleobiological interpretation was only possible after the discovery of a fairly complete skeleton of C. bravoi, and by using an enlarged sample of Murinae for comparison, including some rat-like species and insular species such as the native Philippine Cloud rats Phloeomys and Crateromys (Oliver et al., 1993). Phloeomys and Crateromys are large arboreal, slow-moving rodents characterized by a diet based on vegetal matter (Oliver et al., 1993 and Schauenberg, 1978). In this article, we seek answers to the following questions: Does the semi-complete skeleton of C. bravoi indicate a rat-like murine? Does this skeleton indicate a specialized mode of locomotion, or adaptation to a wide range of environmental conditions?
Material and methods
An almost complete but disarticulated skeleton of C.bravoi (TFMCVF872; Museo de la Naturaleza y el Hombre, Santa Cruz) was found in 1982 by J.-J. Hernández Pacheco in the Cueva del Viento, a lava tube in the North of Tenerife, which yielded Holocene remains (Michaux et al., 1996). It consists of all the postcranial bones except ribs, breast bone, collar bones, hyoid bone, and knee caps. Mandibles and skull of this specimen were not well preserved, so a complete cranium of C. bravoi (TFMCVF873; Museo de la Naturaleza y el Hombre, Santa Cruz) was added for the study together with a cranium of C.tamarani (ZFMK 2010.308; Zoologisches Forschungsmuseum Alexander Koenig, Bonn) in order to complement the data of López-Martínez and López-Jurado (1987). Skulls and skeletons of several Old World rats were used for comparison (Muséum national d’histoire naturelle, Paris): two species that evolved in isolation, Phloeomys cumingi (Waterhouse, 1839) and Crateromys schadenbergi (Meyer, 1895), and several continental species such as the rather large Sundamys muelleri (Jentink, 1879), Rattus rattus (Linnaeus, 1758), Rattus norvegicus (Berkenhout, 1769), Mus musculus (Linnaeus, 1758), and Arvicanthis niloticus (Geoffroy, 1803).
Two approaches have been followed. First, in addition to the semi-complete C.bravoi skeleton, skeletons of P.cumingi, R.norvegicus, and skulls of C. bravoi, C. tamarani, P. cumingi, R. rattus were scanned using a computerized microtomography (μCT) scanner. Pixel size and slice thickness were adjusted according to specimen size. Voxel size ranged from 79.78 μm for the sub-complete skeleton of C. bravoi to 37.6 μm for the other specimens. Following volume data segmentation with Avizo 6.3 (VSG) via thresholding, 3D surfaces representing the cranium and the bones of these specimens were reconstructed, on which measurements were taken. Second, measurements were taken from digital photos of skulls and skeletons of a sample of eight species (MNHN Paris collections): P.cumingi, C.schadenbergi, S.muelleri, A.niloticus, M.insularis, R.rattus, M.musculus. Measurements (Fig. 1), which allowed us to estimate the size and body mass of C.bravoi, as well as several morphofunctional indices in relation to its locomotion mode, following Samuels and Van Valkenburgh (2008). For C. tamarani and M. insularis measurements were obtained from López-Martínez and López-Jurado (1987) and Hutterer et al. (1988), respectively.
The measurements are defined as follows for limbs and girdles: SL: Scapula Length; PL: Pelvis Length; HEB: Epicondylar Breadth of the Humerus; HL: Humerus length; HLMD: Midshaft mediolateral Diameter of the Humerus; DPCL: Length of the Deltopectoral Crest; UMLD: Midshaft mediolateral Diameter of the Ulna; RL: Radius length; FUL: Functional Length of the Ulna; UOL: Length of Olecranon Process; FADP: Midshaft Anteroposterior Diameter of the Femur; FEB: Femur Distal Epicondylar Breadth; FGT: Height of the Greater Trochanter of the Femur; FL: Femur Length; TL: Tibia Length; TMLD: Midshaft Mediolateral Diameter of the Tibia; TSL: Tibial tuberosity Length; MC3L: Metacarpal 3 Maximum Length; Mph3p: Maximum Length of Proximal phalanx 3; Mph3t: Maximum Length of Phalanx 3 of the forelimb; MT3L: Maximum Length of Metatarsal 3; PPh3t: Maximum Length of Distal Phalanx 3. A few cranial measurements were also used to infer body mass (LTRL: Lower Tooth Row Length; LTRW: Lower Tooth Row Width, IMD: Mesiodistal Diameter of the Upper Incisor at wear level), and size of eye sockets that may indicate the orientation of the eyes (IOW, Interorbitary Width; ZW, zygomatic Width).
Morphofunctional indices follow Samuels and Van Valkenburgh (2008): SMI: Shoulder moment index (DPCL/HL); BI: Brachial index (RL/HL); HRI: Humeral robustness index (HMD/HL); HEB: Humeral epicondylar index (HEB/HL); OLI: Olecranon length index (UOLF/FUL); URI: Ulnar robustness index (UMLD/FUL); CI: Crural index (TL/FL); FRI: Femoral robustness index (FADP/FL); GI: Gluteal index (FGT/FL); FEB: Femoral epicondylar index (FEB/FL); TRI: Tibial robustness index (TMLD/TL); TSI: Tibial spine index (TSL/TL); PES: Index of relative length of the foot (MT3L/FL); IM: Intermembral index (HL + RL)/(FL + TL); MANUS: Manus digit 3 proximal phalanx length divided by metacarpal 3 length (Mph3p/MC3L). It indicates the relative proportions of proximal and distal elements of the manus and size of the palmar surface; CLAW: Manus digit 3 terminal phalanx length divided by pes digit 3 terminal phalanx length (Mph3t/Pph3t). It indicates the size of manual claws relative to the pedal claws.
Statistical procedures were performed with Statistica (version 6.0). The morphological variability of extant and extinct rodents was quantified using Samuels and Van Valkenburgh's (2008) procedure based on measurements of morphological (e.g. osteological and muscular) characteristics used to compute indices. For each analysis, a set of 13 robustness, morphofunctional and proportion indices were considered: SMI, BI, HRI, HEB, OLI, URI, CI, GI, FRI, FEB, TRI, TSI, and IM. Manova, in association with a test of significance (Wilk's Lambda test), was performed on these indices in order to assess the effects of life style. A factorial discriminant analyses (DA) was used to infer the mode of locomotion of C.bravoi (Fig. 2). Variables used for running these analyses were first the postcranial indices provided by limb measurements, groups being assemblages of taxa of known mode of locomotion. The database provided by Samuels and Van Valkenburgh (2008) was used following Bover et al. (2010). This database included 37 species of rodents distributed into seven categories of locomotion: terrestrial, semiaquatic, arboreal, fossorial, semifossorial, gliding, and richochetal. The postcranial indices of the complete specimens of the extant P.cumingi and M.musculus, and the extinct C. bravoi were integrated into the dataset as ungrouped cases. The number of measured individuals per species is variable and occasionally very low. A specimen of R.norvegicus was also measured in order to estimate the reliability of the measurements compared to the results of Samuels and Van Valkenburgh (2008).
Allometric relationships known between the proportions of the skeleton and the body mass in rodents allowed for an estimation of the body mass of C.bravoi and C. tamarani following the formulas given by Parra and Jaeger (1998) and Hopkins (2008). Different measurements were considered: IMD (Mesiodistal Diameter of the Upper Incisor at wear level) (Parra and Jaeger, 1998), LTRL (Lower Tooth Row Length) and the RTRA or Rectangular Lower Tooth Row Area (LTRL × LTRW) (Hopkins, 2008).
Results
Fig. 3 presents the 3D reconstruction of the skeleton of C.bravoi based on the semi-complete skeleton TFMCVL782, except the skull, which is from specimen TFMCVL783, and Fig. 1 and Fig. 4 illustrate several postcranial bones from the semi-complete skeleton.
Body size
Estimated head and body length of C.bravoi is 227 mm and tail length at least 157 mm. Values for C. tamarani were respectively 287 mm and 200 mm (López-Martínez and López-Jurado, 1987). For P.cumingi the head and body length is between 400–435 mm, and tail length between 270 and 315 mm (Nowak, 1999). Relative tail length can be estimated through the number and the length of caudal vertebrae depends on the completeness of the fossilized specimen. C.bravoi had at least 21 caudal vertebrae while C. tamarani had about 32 such vertebrae (López-Martínez and López-Jurado, 1987).
Body mass
Hopkins’ approach (2008) provided several estimations of the body mass of C.bravoi. LTRL (Lower Tooth Row Length) gives an estimation ca 830 g for the specimen TFMCVL783, ten species of Murinae being used as reference. The area of the lower dental row (LTRL × LTRW) provides a body mass of ca. 1020 g taking the same reference species. Taking 43 species of muroids with a body mass below 5 kg, the estimated body mass is ca 930 g, and with 70 species of rodents under 5 kg, 890 g. According to these same regressions, the body mass of the studied individual is estimated as between 758 and 871 g. The body mass of C. tamarani is ca 824 g based on LTRL (López-Martínez and López-Jurado, 1987). The Mesiodistal Diameter of the Upper Incisor at wear level (IMD, Parra and Jaeger, 1998) yields a body mass of 797 g for TFMCVL783, and 633 g for the studied nearly complete specimen TFMCVL782. These results are below the estimates previously given by Michaux et al. (1996).
Body shape
Forelimbs are shorter in C.bravoi than in P.cumingi and R.norvegicus (IM). Index EP indicates that R. norvegicus has longer feet than P. cumingi and C. bravoi, values in C. bravoi being intermediate. MANUS index shows that hands have a larger plantar area in P. cumingi than in C. bravoi and R. norvegicus. CLAW index shows that foot claws are stronger than hand claws in C. bravoi and R. norvegicus. C.bravoi shows high values of robustness indices (HRI, URI, FRI, and TRI) compared to P. cumingi, R. norvegicus and R. rattus: values are greater, except FRI, and R. norvegicus and R. rattus are close to C. bravoi.
The zeugopodes and stylopodes of C. bravoi are robust. Indices characterizing forelimbs (SMI, HEB, and OLI) show that C. bravoi differs by a stronger deltoid crest (Fig. 4: Image 5a, b), a more powerful humerus epicondyle, and a shorter olecranion process relative to ulna length compared to the three species used as reference. For indices in relation to hind limbs (GI, FEB, and TSI), all the highest values but FEB (relative width of the femoral epicondyle) are displayed by P. cumingi. The FEB value in C. bravoi is slightly higher than in P. cumingi. The tibial tuberosity (TSI) is relatively less developed in C. bravoi than in P. cumingi and R. norvegicus.
The ratio SL/PL distinguishes C.bravoi from P.cumingi and R.norvegicus, the scapula being less developed in C. bravoi; only the acromion of the latter's scapula is stronger compared to R. norvegicus (Fig. 4: 2). However P. cumingi has more developed girdles in relation to the size of the skull (SL/CBL, PL/CBL). The stylopod of P. cumingi is longer than its zeugopod (BI and IC).
Morphofunctional indices
Compared to Samuels and Van Valkenburgh (2008), we found very similar results for the specimens of R.norvegicus that we measured, indicating the reliability of our measurements. Noteworthy observations are: High robustness indices (HRI, URI, FRI, TRI), the rather weak MANUS (palmar area), and the low IM calculated for C.bravoi exclude the fossorial, gliding, semiaquatic, and richochetal locomotion categories (Samuels and Van Valkenburgh, 2008). It is very unlikely that C. bravoi belongs to the first two locomotion categories; high FRI, HRI, BI, CI, and PES indices exclude any attribution to the richochetal mode of life. Several indices in C. bravoi (OLI, CLAW, FRI, FEB, and EP) are close to indices observed for tree-dwelling rodents according to Samuels and Van Valkenburgh (2008). However, BI and MANUS indices do not indicate a long humerus and hands as expected for such rodents. Robust limbs (HRI, URI, and FRI), with long and strong muscles (SMI, HEB, GI, FEB, TSI) could also indicate a semifossorial rodent but C. bravoi displays short forelimbs (high BI), rather weak claws, and its limbs are of a different length (IM = 0.7). C.bravoi has a weak olecranon process, low value of OLI characterizing forelimbs not clearly modified for digging. BI and MANUS indices are close to the mean values for a terrestrial locomotion (Samuels and Van Valkenburgh, 2008). Nevertheless many rodents, richochetal included, show some abilities in digging.
A discriminant analysis (Fig. 2) following Samuels and Van Valkenburgh (2008), with the addition of the P.cumingi (arboreal), M.musculus (terrestrial), the extinct taxon as an unclassified case, and R.norvegicus to test the reliability of our measurements, allowed their discrimination among groups of different locomotory habits. Morphologies in relation with these locomotory habits could be separated on the first three discriminant axes that accounted for 51.9%, 27.6%, and 13.9%, respectively. MANOVAS on robustness, morphofunctional and proportion indices indicated a significant differentiation of the morphology within the dataset involving the different locomotory habits (Wilk's Lambda test: Value = 0.00259, F = 6.891, P < 0.001). LD1 mainly distinguished ricochetal rodents from all other lifestyles, LD2 tended to discriminate gliding and fossorial taxa, and LD3 mainly separated semiaquatic ones. C.bravoi was included as an ungrouped case in the classification phases. This analysis shows that C. bravoi is closer to Phloeomys (P. cumingi and P. pallidus) than to Rattus (R. norvegicus and R. rattus), and seems to show morphological similarities with arboreal, semifossorial, and terrestrial rodents (Fig. 2). On the one hand, C. bravoi appears to be closer to the semifossorial rodents than arboreal forms on the first three discriminant axes (Fig. 2). On the other hand, and according to statistical analyses performed on postcranial indices, a posteriori probabilities (Appendix A) classify C. bravoi as an arboreal rodent first (86.9%) and then as a semifossorial to (12.8%). This classification showed 91.2% correct classification of species. Fossorial, gliding and ricochetal groups showed 100% correct classification while all other groups had less than 90% correct classification but more than 75%: 88.9% for semifossorial, 87.5% for semiaquatic, 81.8% for arboreal and 85.7 terrestrial. As previously noticed by Bover et al. (2010), most misclassifications were about arboreal and terrestrial taxa, their skeletal proportions remaining more variable. Mahalanobis distances (Appendix B) also indicate that the morphological features of Canariomys are very close to that of arboreal and semifossorial groups (d [C. bravoi s-arboreal] = 23.62, and d [C. bravoi-semifossorial] = 27.04).
Discussion
Skeletons provide insight into the size of a mammal and its mode of life. The appendicular skeleton informs about locomotion habits, whereas the skull provides some information on the diet and the relative development of sense organs. The following discussion mainly focuses on C.bravoi but also involves C. tamarani as it offers another example of insular evolution that may resemble that of C. bravoi.
Before discussing the results, it is worth mentioning briefly the role of allometry. Using the same dataset, Samuels and Van Valkenburgh (2008) analyzed the influence of allometry on the postcranial morphology by performing regressions of log-transformed measurements against the log of the Geometric Mean (GM corresponds to a measure of body size; see Samuels and Van Valkenburgh (2008) for a complete description of the method). They found that most of the characters showed no significant deviation from isometry. Only the Femoral epicondylar breadth (FEB) and the metatarsal three length (MTL3L) showed a significant positive and negative allometry respectively (Samuels and Van Valkenburgh, 2008, Table 4 p. 1390). This virtual absence of interspecific allometry seems to match the pattern of geometric similarity described for small mammals (Biewener, 1990). In small mammals such as rodents, changes in limb posture and muscle insertion areas rather than strong skeletal allometry should compensate for an increase in size (Samuels and Van Valkenburgh, 2008), a situation that likely prevailed in our analysis. Fig. 4 shows that C.bravoi differs from R.norvegicus by stronger humerus and caudal vertebrae, while Phloeomys has a more slender humerus and a much longer acromion, this process being longer in C. bravoi than in R. norvegicus.
The rather similar head to body proportions in C.bravoi, C. tamarani, and P.cumingi suggest that the tails of both Canariomys species had probably the same balancing function as in P. cumingi, a result that is in agreement with the ability of climbing trees.
C.bravoi differed from continental rats by its relative large size, tail length, and body proportions. Its long hind limbs give it a squat posture as seen in other tree-dwelling mammals, as proposed by the reconstruction of its skeleton (Fig. 3). Feet longer than hands also suggest an intermediate body shape between rats and arboreal murine like Phloeomys. Postcranial robustness indices HRI, URI, and TRI show that C. bravoi was able to climb in trees. On the other hand, indices TSI, SMI, GI, and HEB support a semifossorial mode of life. Index FEB tends to locate C.bravoi close to tree-dwelling rodents, while its OLI index seems only compatible with a modest digging capacity. CLAW and IM indices indicate a rather weak development of claws on the forelimbs. Claws of similar length on the hands and feet, and rather long posterior limbs are compatible with climbing on trees and walking on branches. C.bravoi was a strong and powerfully muscled rat able to move on different substrates from the ground floor up to trees, also showing some skills for digging, at least for scratching the ground for food.
Evolution of mammals on islands results in the origination of species that may strongly differ from their continental relatives. P.cumingi, which is restricted to the island of Luzon (Philippines), shows many characteristics of an insular mammal (Musser and Heaney, 1992): large size with unusual body proportions. However, the comparison with the Canary Archipelago is not straightforward. The Philippines Archipelago is located in a zone of complex geological history, and Phloeomys had to compete repeatedly with newcomers. The Phloeomys lineage had individualized earlier than the Canariomys lineages. The molecular phylogeny provided by Lecompte et al. (2008) put different lineages of Murinae into a time sequence, the Phloeomys and Crateromys lineage being the oldest, followed by the Rattus/Sundamys group and the Arvicanthini. Compared with Phloeomys, the two species of Canariomys have experienced a different situation by evolving during a shorter time, without any mammalian competitor, along with an herbivorous giant lizard. No snake occurs in the Canary Islands today, and the only hint of a former existence is a single vertebra from Upper Miocene sediments of Lanzarote (Barahona et al., 1998). Each Canariomys simply had to exploit available resources, a situation that could explain the weak differentiation of their skeletal features compared to continental rats. In contrast, Phloeomys evolved in a competitive context and in the presence of an array of predators such as carnivores and wild pigs (Heaney et al., 2010), or large snakes (Taylor, 1922), and as a response might have had to develop a specialized arboreal life style.
Tenerife had until recently an important forest cover with the exception of xerophilous vegetation in the dry lower part of the island, and poor vegetation above 1800 m (Del-Arco, 2008 and Del-Arco et al., 2006). Such a situation has been roughly maintained since the Pliocene. Prior to the Pliocene, warm waters prevailed around the archipelago, and the flora was sub-tropical to tropical (Meco et al., 2010 and Schmincke, 1968). Many members of this flora survive today where altitude and winds maintain a rather high humidity all year long. The laurel forests as well as the lowland plant communities of Tenerife and Gran Canaria harbour many local tree species with edible fruits, such as Apollonias barbujana, Arbutus canariensis, Laurus azorica, Myrica faya, Ocotea foetida, Persea indica, Phoenix canariensis, Pinus canariensis, Pistacia lentiscus, Viburnum tinus, or Visnea mocanera (Kunkel, 1982). Introduced Ship rats (R.rattus) have invaded the laurel forest of all islands and feed principally on these fleshy fruits, with a preference for smaller fruits (Delgado Garcia, 2000). According to size differences, this suggests that the extinct Canariomys was able to feed regularly on larger fruits, a conclusion in accordance with microwear analyses (Firmat et al., 2010).
The still unknown ancestor of C.bravoi found in Tenerife a wide range of environmental conditions, among which forests and shrubs offered protection against birds of prey. The somewhat reduced orbital cavity in the two Canariomys species may consequently mirror the modification observed in Phloeomys. However, causes for evolving such an association of morphological features were different: on the one hand, a protective cover and the lack of terrestrial predators, on the other hand, a shift toward forested refuge zones.
Once successfully settled, the ancestor of C. bravoi had to face strong intra-specific competition, and suffered from exceptional environmental contingencies such as volcanic eruptions or brief climatic events. The rat-sized murine of African origin evolved into a heavily built rat whose skeleton had little in common with that of a big rat. The changes observed in the proportions of its limbs indicate that C. bravoi could perform a rather wide spectrum of activities on the floor as well as in trees, including some skills for digging. This latter possibility can be a side effect of size increase and adaptation to a more vegetation-based diet. Search for fallback foods under harsh environmental conditions may have played a role in the evolution of C. bravoi, a result that mirrors the interpretation of its dental microwear patterns (Firmat et al., 2010). It is worth mentioning here that some semifossorial skills were also assumed for the extinct dormouse of the Balearic Islands: Hypnomys is considered to have had a wider niche than its continental ancestor Eliomys (Bover et al., 2010 and Hautier et al., 2009). Similarly, the extinct Murinae Mikrotia of the Gargano Peninsula (Parra et al., 1999) is interpreted as semifossorial. The digging skills might be in relation to the widening of the niche under isolation.
Acknowledgments
This note was written as a tribute to Professor Nieves Lopéz-Martínez, our friend, companion in the field, and leading Spanish palaeontologist who died in 2010. She was part of our research team and, as such, contributed to the revival of studies of the island fauna of the Canary Islands.
Lionel Hautier gratefully acknowledges Sidney Sussex College (Cambridge, United Kingdom) and a research grant from the Leverhulme Trust (United Kingdom). Dr Christiane Denys gave access to the MNHN collections, Suzanne Jiquel and Anne-lise Charruault (ISEM, Montpellier) helped. Microtomographic images of the semi-complete skeleton were acquired at the Centre de Microtomographie of University of Poitiers. The present work developed results of a preliminary study initiated by Mikael Antocio during the academic year 2010–2011, at the Paleontology Department, University of Montpellier-2. Supplementary data can be obtained from J. Michaux (jjmichau@univ-montp2.fr). Publication ISEM–UMR 5554 no 2012-080.
Post-probabilities calculated from the discriminant analysis performed on postcranial indices.
AtocheP.PazJ.A.RamírezM.A.OrtizM.E.Evidencias arqueológicas del mundo romano en Lanzarote, (Islas Canarias)1995Servicio de Publicaciones del Cabildo de LanzaroteArrecifep. 149BarahonaF.RageJ.C.García-TalaveraF.The first record of snakes on the Canary Islands: a vertebra from the Upper Miocene of LanzaroteAmphibia-Reptilia191998419–425BiewenerA.Biomechanics of mammalian terrestrial locomotionScience25019901097–1103BocherensH.MichauxJ.BilliouD.CastanetJ.García TalaveraF.Contribution of collagen stable isotope biogeochemistry to the paleobiology of extinct endemic vertebrates from Tenerife (Canary Islands, Spain)Isotopes Environ. Health Stud.392003197–210BocherensH.MichauxJ.García TalaveraF.Van der PlichtJ.Extinction of endemic vertebrates on islands: the case of the giant rat Canariomys bravoi (Mammalia, Rodentia) on Tenerife (Canary Islands, Spain)C. R. Palevol.52006885–891BoverP.AlcoverJ.A.MichauxJ.HautierL.HuttererR.Body shape and life style of the extinct Balearic dormouse Hypnomys (Rodentia, Gliridae): new evidence from the study of associated skeletonsPLoS ONE5122010e1581710.1371/journal.pone.0015817/aCrusafont-PairoM.PetterF.Un muriné géant fossile des îles Canaries: Canariomys bravoi gen. nov., sp. nov. (Rongeurs, Muridés)Mammalia281964608–611Del-Arco, M., 2008, Consecuencias del cambio climático sobre la flora y vegetación canaria. In: Méndez Pérez, J.M., M. Vázquez Abeledo (Eds.), El Cambio Climático en Canarias. Academia Canaria de Ciencias. Serie Monografías, 1, pp. 79–100.Del-ArcoM.Pérez-de-PazP.L.AcebesJ.R.González-ManceboJ.M.Reyes-BetancortJ.A.BermejoJ.A.de-ArmasS.González-GonzálezR.Bioclimatology and climatophilous vegetation of Tenerife (Canary islands)Ann. Bot. Fennici432006167–192Delgado GarciaJ.D.Selection and treatment of fleshy fruits by the ship rat (Rattus rattus L.) in the Canarian laurel forestMammalia64200011–18DubeyS.KoyasuK.ParapanovR.RibiM.HuttererR.VogelP.Molecular phylogenetics reveals Messinian, Pliocene, and Pleistocene colonizations of islands by North African shrewsMol. Phyl. Evol.472008877–882FirmatC.Gomes RodriguesH.RenaudS.ClaudeJ.HuttererR.García-TalaveraF.MichauxJ.Mandible morphology, dental microwear, and diet of the extinct giant rats (Canariomys, Rodentia: Murinae) of the Canary Islands (Spain)Biol. J. Linn. Soc.101201028–40HautierL.BoverP.AlcoverJ.A.MichauxJ.HautierL.BoverP.AlcoverJ.A.MichauxJ.Mandible morphometrics, dental microwear pattern, and palaeobiology of the extinct Balearic Dormouse Hypnomys morpheusActa Palaeontol. Pol.542009181–194Heaney, L.R., Dolar, M.L., Balete, D.S., Esselstyn, J.A., Rickart, E.A., Sedlock, J.L., 2010. Synopsis of Philippine Mammals. Field Museum website: http://www.fieldmuseum.org/philippine_mammals/.HopkinsS.S.Ressessing the mass of exceptionally large rodents using tooyth row length and area as proxies for body massJ. Mammal.892008232–243HuttererR.García TalaveraF.López MartínezN.MichauxJ.New chelonian eggs from the Tertiary of Lanzarote and Fuerteventura, and a review of fossil tortoises of the Canary Islands (Reptilia, Testudinidae)Viearea261998139–161HuttererR.López-JuradoL.F.VogelP.The shrews of the eastern Canary Islands: a new species (Mammalia: Soricidae)J. Nat. Hist.2119871347–1357HuttererR.López-MartínezN.MichauxJ.A new rodent from Quaternary deposits of the Canary Islands and its relationships with Neogene and recent murids of Europe and AfricaPalaeovertebrata81988241–262KunkelG.Die Kanarischen Inseln und ihre Pflanzenwelt1982G. FischerStuttgart, New York202 pLecompteE.AplinK.DenysC.CatzeflisF.ChadesM.ChevretP.Phylogeny and biogeography of African Murinae based on mitochondrial and nuclear gene sequences, with a new tribal classification of the subfamilyBMC Evol. Biol.82008199López-JuradoL.F.López-MartínezN.Presencia de la rata gigante extinguida de Gran Canaria (Canariomys tamarani) en una cueva de habitación aborigenEl Museo Canario48199119–22López-MartínezN.López-JuradoL.F.Un nuevo múrido gigante del Cuartenario de Gran Canaria Canariomys tamarani nov. sp. (Rodentia, Mammalia). Doñana Acta VertebrataPubl. Ocas.219871–66MecoJ.MuhsD.R.FontugneM.RamosA.J.G.LomoschitzA.PattersonD.Late Pliocene and Quaternary Eurasian locust infestations in the Canary ArchipelagoLethaia201010.1111/j.1502-3931.2010.00255.xMichauxJ.López-MartínezN.Hernández-PachecoJ.J.A 14C dating of Canariomys bravoi (Mammalia, Rodentia), the extinct giant rat from Tenerife (Canary Islands, Spain), and the recent history of the endemic mammals in the ArchipelagoVie Milieu461996261–266MusserG.G.CarletonM.D.Family MuridaeWilsonD.E.ReederD.A.Mammal Species of the World a Taxonomic and Geographic Reference2005Johns HopkinsBaltimore501–755MusserG.G.HeaneyL.R.Philippine rodents: Definitions of Tarsomys and Limnomys plus a preliminary assessment of phylogenetic patterns among native Philippine murines (Murinae, Muridae)Bull. Amer. Mus. Nat. Hist.21119921–138NavarroJ.F.MartínE.RodríguezA.Las primeras etapas del programa de excavaciones en las Cuevas de San Juan y su aportación a la diacronía de la Prehistoria de CanariasInvestigaciones Arqueológicas en Canarias21990189–201NowakR.Walker's Mammals of the World6th ed.1999Johns HopkinsBaltimore and London1936 pOliverW.L.R.CoxC.R.GonzalesP.C.HeaneyL.R.Cloud rats in the Philippines – Preliminary report on distribution and statusOryx27199341–48ParraV.JaegerJ.J.Estimation de la taille et du poids corporel chez les rongeurs (Rodentia, Mammalia) à patir de la taille des incisivesC. R. Acad. Sci. Paris, Ser. IIa326199879–85ParraV.JaegerJ.J.BocherensH.The skull of Microtia, an extinct burrowing murine rodent of the late Neogene Gargano paleoislandLethaia32199989–100SamuelsJ.X.Van ValkenburghB.Skeletal indicators of locomotor adaptations in living and extinct rodentsJ. Morphol.26920081387–1411SchauenbergP.Note sur le Rat de Cuming Phloeomys cumingi Waterhouse 1839 (Rodentia, Phloeomyidae)Rev. Suisse Zool.851978341–347SchminckeH.U.Pliozäne, subtropische Vegetation auf Gran CanariaNaturwissenschaften551968185–186TaylorE.H.The snakes of the Philippines IslandsPhilipp. Bur. Sci. Manila1922312
Illustrations of Canariomys bravoi skull (dorsal, ventral and lateral views) and mandible (dorsal and lateral views) TFMC873, and measurements taken on skull and postcranial bones based on TFMC872 for bones and TFMC873 for the skull and mandible (see Section 2 for abbreviations).
Illustrations du crâne (vues dorsale, ventrale et latérale) tet de la mandibule (vues dorsale et latérale) de Canariomys bravoi (TFMC873) et mesures prises sur le crâne et les os post-crâniens d’après le spécimen TFMC872 pour les os et TFMC873 pour le crâne et la mandibule (voir Section 2 pour les abréviations).
Factorial discriminant analysis among groups of different locomotor habits. (A) Plot of axes one and two; (B) Plot of axes one and three.
Analyse factorielle discriminante des différents groupes à partir des modes de locomotion. (A), Distribution selon les axes 1 et 2 ; (B), selon les axes 1 et 3.
3D reconstruction of Canariomys bravoi skeleton according to specimens TFMC872 (semi-complete skeleton) and TFMC873 (skull).
Reconstitution 3D du squelette de Canariomys bravoi, d’après les spécimens TFMC872 (squelette sauf crâne) et TFMC873 (crâne).
Scapula, humerus and first caudal vertebrae of Phloeomys cumingi (1, 4, 7; MNHN CG 1882-144), Canariomys bravoi (2, 5, 8; TFMC872) and Rattus norvegicus (3, 6, 9). Each bone scaled to the same size. 1–3: right scapula; 4–6: right humerus (respectively a, medial and b, posterior views); 7–9: first caudal vertebrae (dorsal view).
Omoplate, humérus et premières vertèbres caudales de Phloeomys cumingi (1, 4, 7 ; MNHN CG 1882-144), Canariomys bravoi (2, 5, 8 ; TFMC872) et Rattus norvegicus (3, 6, 9). Chaque os mis à une même taille. 1–3 : omoplate droite ; 4–6 : humérus droit, (respectivement vue médiale et vue postérieure) ; 7–9 : premières vertèbres caudales, (vue dorsale).