Many genera of arvicolid rodents (voles) contain species that spend a considerable amount of time underground. Incisors provide a considerable effort in the digging process and a difference in the enamel microstructure of incisors is expected between species that spend most of their life above ground and those who have greater specializations for a subterranean habitat. The ratio between the thicknesses of the two layers composing the enamel of the incisors has proved to be an effective proxy to infer fossorial activity of extinct forms of Arvicola but this ratio exhibited no phylogenetic signal (although lifestyle was in a greater extent related to phylogeny). We were able to infer ancestral lifestyles (of extinct populations and of nodes of the phylogeny) from enamel morphologies. Inclusion of fossils did not solve the uncertainties associated with the lifestyle for the last common ancestor of extant Arvicola species and it did not change the inferred lifestyle of several other, less inclusive, clades.
De nombreux genres de campagnols (Rodentia, Arvicolinae) incluent des espèces au mode de vie clairement fouisseur et chez lesquelles les incisives jouent un rôle actif lors du fouissement. Cette situation permet d’envisager que la microstructure de l’émail des incisives pourrait révéler des différences entre espèces qui vivent surtout en surface et celles qui ont un mode de vie souterrain marqué. Ce travail montre que le rapport des épaisseurs des deux couches de l’émail des incisives est un bon d’indicateur du caractère fouisseur des formes actuelles et qu’il pourrait s’appliquer aux formes éteintes d’Arvicola. Ce rapport n’est pas corrélé à la phylogénie (bien que le mode de vie soit dans une grande mesure relié à la phylogénie). L’inclusion des populations éteintes n’améliore pas les incertitudes sur les inférences des modes de vie ancestraux pour le dernier ancêtre commun des espèces actuelles d’Arvicola et elle ne change pas les états ancestraux des clades moins inclusifs.
Arvicola, Europe, Enamel microstructure, Lifestyle, Ancestral state reconstruction, Phylogenetic signalArvicola, Europe, Émail, Microstructure, Mode de vie, Reconstruction d’états ancestraux, Signal phylogénétiquepresentedWritten on invitation of the Editorial BoardIntroduction
Paleoclimatic and paleoenvironmental reconstructions inferred from small mammal communities are quite common in paleontology (Chaline et al., 1995, Marquet, 1989 and van Dam, 1997). However, these reconstructions are often limited by the use of communities predominantly composed of taxa having closely related extant representatives, whose environmental preferences are well known. Any insight into extinct taxon lifestyle would be of extreme interest for studies on small fossil mammals. Arvicola is one of the few rodents displaying iterative evolution to a semi-aquatic and fossorial lifestyle, which makes it a very interesting model to test the rhythm and tempo of adaptive phenotypic changes to those special environments. In addition, Arvicola is nested within a clade where these changes have also occurred several times in closely related forms (e.g., in Ondatra, Myodes, Microtus, etc.), which makes lifestyle of fossil Arvicola or pre-Arvicola forms difficult to estimate without using a morphological or geochemical inferential approach.
Burrowing taxa, such as fossorial water voles (Airoldi et al., 1976, Laville, 1989 and Laville et al., 1989), use incisors as picks for digging, whether or not they are the primary digging apparatus. Skull adaptations to the digging activity (Cubo et al., 2006), as well as modifications of incisor enamel microstructure (Stein, 2000), have been documented. However, very little is known about the relationships between enamel microstructure and lifestyle in rodents, though already in the mid 19th century Tomes (1850) showed that rodents have two layers of enamel, in contrast to lagomorphs, which have only one layer. As mammalian enamel in general, rodent incisor enamel is composed by hydroxyapatite crystallites (called enamel prisms) that are organized in a variety of structures that originate at the enamel–dentine side and run through the entire thickness of the tissue to the outer enamel surface. The Radial Enamel forms the external layer (Portio Externa). It is the simplest of the three main types of tooth enamel (prisms are arranged parallel to one another, radiate from the enamel–dentine junction towards the occlusal surface, and intersect the occlusal surface at approximately 90°). The inner layer (Portio Interna) of rodent incisors is usually composed of Hunter-Schreger Bands (the prisms originating at the dentine–enamel border for a given layer are oriented in one way, while those of the surrounding layers are oriented in the other direction) so that, at a high magnification analysis, the surface appears banded, being the bands formed by zones of enamel prisms with different orientations. Enamel is the easiest material to retrieve in the fossil record, due to its high degree of mineralization, and it is rich in information about function and phylogeny (Koenigswald, 1980 and Koenigswald, 2004 and references therein). Many studies (Koenigswald, 2004, Martin, 2005 and Rensberger et al., 1984) have been published about phylogeny of mammal groups inferred from their tooth enamel structure, also known under the name Schmelzmuster (Koenigswald, 1980). Yet, only a few studies are known about the relationship of enamel microstructure to incisor function in rodents. In particular, Vieytes et al. (2007) have shown that in some South American caviomorph rodents belonging to the family Ctenomyidae, which are fossorial, the thickness of the external Radial Enamel portion differs according to genus and to hardness of the soil. In particular, Vieytes et al. (2007) demonstrated that the higher values of External Index of Ctenomys talarum compared with C. australis are consistent with the former digging in harder soils. Likewise, among North American geomyid burrowing rodents, higher External Indices (Buzas-Stephens and Dalquest, 1991) are positively correlated with greater tooth-digging specialization, from the primarily scratch-digger Geomys, in which Radial Enamel represents only 20% of the total thickness, to the relatively more specialized tooth-diggers Cratogeomys, in which it represents 25%, and especially Thomomys, adapted to dig in hard soils and for which the thickness of the Portio Externa reaches up to 40% in the incisors.
The aims of this study were to determine possible differences in the External Index between semi-aquatic and fossorial populations of Arvicola, and if these differences were significant, to use this index as a proxy to infer ecological preferences of extinct forms of this genus. Analyses were performed on eight extant taxa, four fossorial and four semi-aquatic, and on fossil specimens of three extinct populations (one from the Middle and two from the Late Pleistocene). This research constitutes a part of an ongoing project on morphometrics, evolution and phylogeny of extinct and extant representatives of Arvicola. In addition to address issues regarding the lifestyle and phenotypic evolution of Arvicola, we also provide a review of the current knowledge about Arvicola taxonomy and phylogeny in order to give an historical framework for the evolution of lifestyles and incisor enamel evolution in that group.
Phylogeny and taxonomy of <italic>Arvicola</italic>
From the 1960s on, it has been generally considered that Arvicola includes two extant species (see e.g. Reichstein, 1963 and Reichstein, 1982): the south-western water vole, A. sapidus, which is found in the Iberian Peninsula and much of France (Baudoin, 1984, Saucy, 2002 and Ventura, 2007), and the northern water vole, A. terrestris, which extends widely throughout the Palaearctic region (Corbet, 1978). While A. sapidus is a semi-aquatic form, it is considered that A. terrestris presents two ecological forms: semi-aquatic and fossorial (Kratochvíl, 1983). Musser and Carleton (2005), following Panteleyev (2001) and Zagorodnyuk (2001) reviews on the genus Arvicola, have demonstrated that the two long established ecological forms of A. terrestris are different taxonomic units, reserving the name of A. amphibius (European water vole) for the semi-aquatic populations and A. scherman (montane water vole) for the fossorial ones. Nevertheless, positive results on the hybridization between terrestris and scherman (Bernard, 1961 and Müller-Böhme, 1936) and certain molecular studies (Piertney et al., 2005 and Taberlet et al., 1998) do not support this taxonomic configuration. Additionally, although the specific name A. amphibius has nomenclatural priority over A. terrestris (Brink et al., 1967 and Musser and Carleton, 2005), the latter seems to have been maintained to avoid “confusion and ambiguity” (see Corbet, 1978). In spite of these controversies we have followed in this study the new taxonomic proposal by Musser and Carleton (2005) in order to contrast it with the results obtained here and give new insights onto this taxonomic problem. The extinct representatives of A. amphibius, which were widespread in Europe from the Middle to the Late Pleistocene, have been traditionally considered semi-aquatic as well as its possible ancestor A. mosbachensis. Such a statement, however, is uncertain and the ancestor of A. sapidus, A. scherman, and A. amphibius may also have been fossorial since reverse evolution is attested at least for A. amphibius.Maul et al. (1998) considered A. cantianus as a synonym of A. mosbachensis. However, the type material is highly fragmentary and we prefer to consider Arvicola cantianus as a nomen dubium. This name should therefore not be employed for the taxonomy of fossil specimens.
For extant taxa, Taberlet et al. (1998) found evidence for a sister group relationship among the Italian A. amphibius populations (A. a. italicus) and all other European populations belonging to A. amphibius and A. scherman. Later on, Piertney et al. (2005) showed that European populations are structured into two clades: A. amphibius korabensis, A. amphibius terrestris and A. amphibius amphibius on the one hand, and A. scherman cantabriae, A. scherman monticola, A. scherman exitus, A. scherman scherman and A. amphibius reta on the other hand. These results strongly suggest that A. amphibius, as currently recognized, is not monophyletic. The phylogenetic structure of the last group of populations may partly reflect the northward colonization routes of these populations after the end of Marine Isotope Stage (MIS) 2 (approximately after 11 ky BP).
We followed Cuenca-Bescós et al. (2001) for the 400 ky BP divergence time between A. sapidus and the other Arvicola species. Recently, Centeno-Cuadros et al. (2009), based on a molecular clock, dated the same split at about 250 ky Bp. This result, however, needs a revision because there is paleontological evidence of A. sapidus in the early Middle Pleistocene of Iberian Peninsula.
Maul et al. (1998) and Masini et al. (2003) suggested that the divergence between “peninsular Italian arvicolas” (=A. amphibius musignani, according to recent systematics, Cagnin, 2008) and the other clade including Arvicola scherman + Arvicola amphibius occurred between 250 and 200 ky BP. No data exist to set the timing for the divergence node between A. scherman and A. amphibius (excluding Italian populations); for this reason we considered the results from Maul et al. (1998) that place the differentiation between Arvicola amphibius terrestris and the European Arvicola scherman at about 125 ky BP. They based this conclusion on the Schmeltzband Differentierungs Quotient (SDQ) ratio, the ratio between enamel thickness measured in the posterior and anterior cutting edges of molar triangles. This age constitutes a lower age limit, but we cannot exclude younger divergence age for this node.
Three extinct populations were placed in the phylogenetic tree according to recent phylogenies and according to their skull and molar morphologies: Arvicola mosbachensis from Mosbach (West Germany) and dated at 500Ka BP (Maul et al., 2000), fossil Arvicola amphibius from Grotta Breuil (central Italy) dated at about 34Ka BP (Alhaique et al., 1998 and Kotsakis, 1991) and fossil Arvicola scherman coming from Bouziès (southern France) dated at about 17ka BP (Aguilar et al., 2008). Arnicola mosbachensis is generally considered to be the sister-group of all other species of Arvicola. Since Signor Lipps effects (sampling bias in the fossil record) may have occurred (Signor and Lipps, 1982), we opted for two strategies concerning the initial splits in Arvicola phylogeny. In a first attempt, the split between A. mosbachensis and all other Arvicola species was dated 100 Ky before the first occurrence A. mosbachensis, that is 605 Ky. This phylogenetic scenario (phylogenetic scenario 1) is possible because the first occurrence of Arvicola sp. is from la Isernia La Pineta in central Italy (Sala, 1996). The second scenario considered that A. mosbachensis was the ancestor of all other Arvicola species (therefore, it is considered as not divergent from that ancestor). The two scenarios are presented in Fig. 2. Based on skull and teeth morphology, the population of Arvicola from Bouziès (France) referred here to A. scherman, was placed as the sister group of all other extant A. scherman, while the Italian extinct population coming from Grotta Breuil and attributed to A. amphibius was considered as the sister taxon of the clade (A. amphibius musignani + A. amphibius italicus).
Materials and methodsMaterial
Our sample includes eight extant and three extinct populations. Lower incisors of thirty two specimens coming from three extant populations of Arvicola amphibius, four extant populations of A. scherman and one extant population of Arvicola sapidus (Fig. 1) were analyzed in order to assess differences, if any, in the external index value (see below). These specimens belong to the following subspecies: A. amphibius terrestris (5 specimens; Sleneset Islands, Norway), A. amphibius musignani (7 specimens, Sila Grande, Italy), A. amphibius italicus (3 specimens; Agno, Switzerland), A. scherman scherman (3 specimens; Prangins, Switzerland), A. s. exitus (3 specimens; Ally, France), A. scherman monticola (3 specimens; Aran Valley, Spain), A. scherman cantabriae (3 specimens; Ribadesella, Spain), A. sapidus (5 specimens; Ebro River delta, Spain). Moreover, we analyzed the lower incisors of eleven fossil specimens belonging to A. amphibius from Grotta Breuil (3 specimens), A. scherman from Bouziès (3 specimens), and A. mosbachensis from Mosbach (5 specimens).
Enamel External Index (EI)
Teeth have been first cut at half of their length, then ground to expose a flat surface and polished. Specimens were then dehydrated with alcohol and etched for 4–5 s with 10% hydrochloric acid to accentuate topographic detail. Afterwards they were sputter-coated with graphite. Photographs were taken using a Philips Scanning Electron Microscope. Total enamel thickness (ET) and Radial enamel thickness or Portio Externa (PE) have been measured, according to Vieytes et al. (2007). An External Index (EI) has been calculated as the ratio between PE and ET (Vieytes et al., 2007) and averaged for each species. Prior to making a comparative analysis of EI values and lifestyle we opted to check whether there was a phylogenetic signal in characters and in that case to use ancestral character reconstruction in order to orient subsequent comparative analyses.
Reconstruction of ancestral states and phylogenetic signal
The phylogeny defined above and presented in Fig. 2 has been used to estimate ancestral state values of Enamel Index as well as for lifestyle for the root and inner nodes of the tree. Lifestyle was considered as varying between two extremes: fossorial vs. semi-aquatic. In order to appraise intermediary lifestyle, we attributed a value of 0 for an exclusively semi-aquatic population and a value of 1 for an exclusively fossorial population. We considered all extant species as exclusive fossorial or semi-aquatic, although we are aware that it approximates the reality, and that semi-aquatic forms can perhaps spend time digging burrows using their teeth. The two behaviours appear not to be mutually exclusive; we therefore considered possible continuous variation between these two extremes. Estimation of ancestral character states were obtained first only using extant taxa, since lifestyles of extinct species cannot be observed. Prior to performing estimations of ancestral values, we tested several evolutionary models in order to assess the best model for character evolution. Possible models are Brownian motion model where the variance of a character is proportional to branch length; Ornstein-Uhlenbeck, that fits a model of random walk with a central tendency with a strength proportional to a given parameter alpha; Pagel's lambda multiplies all internal branches of the tree by a given parameter lambda, leaving tip branches as their original length; Pagel's kappa: it raises all branch lengths to the power kappa; when kappa approaches zero, the model becomes speciational; Pagel's delta; raises all node depths to the power delta; if delta is less than one, evolution is concentrated early in the tree; delta greater than 1 concentrates evolution towards the tips; ACDC models: set by the “eb” parameter, it fits a model where the rate of evolution increases or decreases exponentially through time, under the model r(t) = rho × exp(r × t), where ro is the initial rate and r is the rate change parameter; white noise model: no phylogenetic signal.
The “GEIGER” package for R (Harmon et al., 2008) has been used to estimate the best fit among various model using Akaike Information Criterion (AIC) and the parameters for each model. The original tree was eventually transformed according to the parameters of the best model found. On this transformed tree EI ancestral states were estimated via maximum likelihood procedure using “ape” package for R v. 2.5-3 (Paradis et al., 2004).
Comparative analyses
Results of this previous analysis oriented our selection for a comparative analysis including the phylogeny, transforming the phylogeny, or excluding the phylogeny (if no phylogenetic signal was detected in the traits). In this latter case (no phylogenetic signal), we applied a simple generalized linear model of lifestyle on EI in order to check whether there was an association between the behavioural trait and the morphological one.
ResultsEnamel External Index (EI)
Mean and standard deviation for EI values among populations are shown in Table 1. Mean EI values were below 22 for all semi-aquatic extant species or populations and always above 25 for all living fossorial populations or species (Fig. 3). The range of mean EI values is wider for fossorial (25 to 40) than for semi-aquatic populations (18 to 22). Within populations, variation in EI value can be important, especially for A. scherman cantabriae, which also shows the largest EI values. When all living populations are considered together (semi-aquatic and terrestrial), interpopulation variation is significantly larger than intra-population variation (F(10.33) = 7.3; P < 0.001). However, when lifestyle is taken into account, interspecific variation is significantly larger than intraspecific variation only for fossorial populations (F(3.8) = 11.1; P < 0.005) and no longer in semi-aquatic populations (F(3.16) = 0.3, P = 0.79). Moreover, intrapopulation variation was found to be similar between semi-aquatic and fossorial populations (F(16.8) = 1.21, P = 0.8), while interpopulation variation was greater for fossorial populations (F(3.3) = 26.7, P = 0.011). If we consider population as a random effect, the variation of EI values is significantly explained by the lifestyle (F(1.6) = 7.85, P = 0.03). The EI value of the A. scherman extinct population from Bouziès (Fig. 4A) equals to 35.97; therefore it is similar to that observed in the fossorial A. s. cantabriae. In contrast, the A. amphibius extinct population from Grotta Breuil has a very low EI (18.87) similar to the lowest values among extant semi-aquatic populations. However, the Grotta Breuil specimens are slightly eroded in their external layer and the real value could be slightly larger. Finally, EI value of the extinct population A. mosbachensis (the oldest population), equals 23.66. A. mosbachensis (Fig. 4B) has therefore an EI value intermediate between semi-aquatic and fossorial populations (Table 1); it should also be noted that for this taxonomic unit, EI values were variable among individuals (ranging from 18.43 to 33.96).
Ancestral states reconstructions.
The AIC comparison between the different models of character evolution demonstrated that the White Noise model was the best for EI evolution along the phylogeny including only extant forms. This suggests that there is no phylogenetic signal in EI (that this trait is very labile). The same result was found when extinct forms were included using the first phylogenetic scenario; on the other hand, the best model was Pagel's lambda with lambda equals 0.160 for the second phylogenetic scenario. Since Pagel's lambda is low, this model shows that most of the character evolution is concentrated on terminal branches rather than on internal branches (i.e., that we are not far from a White Noise model). The result obtained with the second scenario can also be interpreted as an artifact coming from errors concerning the divergence time of A. mosbachensis from other Arvicola species. Evolution of lifestyle followed a Brownian Motion model in the tree with only extant taxa.
Comparative analyses
The inference for extinct lifestyle can be made only using a model based on extant taxa. No phylogenetic signal was found for the EI indices using a phylogeny including only extant taxa or our first phylogenetic scenario including fossils. We therefore tentatively opted to check whether lifestyle could be predicted by EI indices of populations using a generalized linear model (glm). We used a binomial family for the response because living forms were considered as exclusively fossorial or semi-aquatic. The model was highly significant with a residual deviance of nearly 0 (P <0.001). Lifestyles were predicted using this logistic model as corresponding to scores on a zero to one scale (0 corresponding to exclusive semi-aquatic populations, and 1 to exclusive fossorial populations). On this scale, values of 0.958, 1, and 0 were found for populations of A. mosbachensis, A. scherman from Bouziès, and A. amphibius from Grotta Breuil. While Grotta Breuil fossils can be considered as semi-aquatic and Bouziès as fossorial with some confidence, uncertainties remain for A. mosbachensis. Indeed, one should consider that there is a great uncertainty with A. mosbachensis because no living populations were sampled with values between 0.22 and 0.25, the interval that separates semi-aquatic from fossorial populations. There is therefore a large uncertainty on the inflexion of the logistic curve. We therefore also interpolated the lifestyle score of A. mosbachensis using a linear interpolation between the semi-aquatic and fossorial populations that got the closest EI indices of that extinct populations. This gave a 0.567 score for that population, which may better reflect the aforementioned uncertainty. In order to incorporate inferences from fossil specimens in the phylogeny, we used their lifestyle estimates and we predicted ancestral character states for nodes in the phylogenetic tree according to the different scenarios and inferences for A. mosbachensis. The best model using GEIGER package was the brownian model for the first phylogenetic scenario when the value of A. mosbachensis was 0.958, while it was Pagel's delta with delta equal to 2.999 for the second phylogenetic scenario for both values of A. mosbachensis and for the 0.567 value of A. mosbachensis with the first phylogenetic scenario. Ancestral character states for lifestyles are given in Table 2. Results show similarities with the tree obtained only with extant taxa, except for deep nodes. In all the models, the ancestral lifestyle for the deep nodes are highly uncertain while ancestral lifestyles for more recent nodes are consistent between methods and have narrower confidence intervals. The inclusion of fossils would rather favor a more fossorial ancestor for all living populations, but the uncertainties around ancestral character state are too important for clearly supporting this conclusion.
Discussion
Previous phylogeographic studies (Piertney et al., 2005 and Taberlet et al., 1998) allowed us to place our results in a phylogenetic framework (Fig. 2). In this context, Arvicola amphibius (as defined by Zagorodnyuk, 2001) is paraphyletic: the last common ancestor of all subspecies of Arvicola amphibius is also the ancestor of all subspecies of Arvicola scherman (Fig. 2).
When considering only extant forms, no phylogenetic signal was found on EI, suggesting that this feature is highly labile during evolution. However, lifestyle was clearly related to phylogeny showing that this trait may evolve more slowly during evolution. In addition, we have found that despite of the different evolutionary behaviour of both characters, there is a significant relationship between the average EI values and lifestyle, fossorial populations showing significantly higher values than semi-aquatic ones: EI values show an important variation, especially in fossorial populations; in addition, there are differences in EI values among fossorial populations. We suggest that this measure can be taken as a proxy for predicting the fossorial activity of Arvicola. The high degree of variation within fossorial populations and the weak relation with phylogeny suggest that this character has a great phenotypic plasticity; however, this still need to be tested by ecological observations and experimentation.
Considering that there is a relationship between EI and lifestyle at the population level, we used this histological trait to infer the lifestyle of the three fossil populations included in our sample. The sample of Arvicola scherman coming from the fossil-bearing locality of Bouziès (Fig. 4A) has an EI of 35.97, almost as high as that of A. scherman cantabriae (39.62), thus suggesting that these animals were fossorial. Such a hypothesis is sustained also by observations on some cranial fragments recovered in the site showing forwardly directed upper incisors, that further support the idea of a burrowing population and an attribution of the Bouziès population to A. scherman. In contrast, the A. amphibius specimens from the Grotta Breuil have an EI of 18.87, close to the lowest values of living semi-aquatic populations, suggesting a semi-aquatic lifestyle. The EI value of the oldest fossil population of our sample (Fig. 4B), coming from the Middle Pleistocene site of Mosbach, is 23.66 and its lifestyle is more difficult to determine because this population is intermediate between the upper limit for semi-aquatic population and the lower limit for fossorial populations. Since our sampling of Arvicola is small at the population level, there are great uncertainties concerning the lifestyle of the Mosbach population. These uncertainties are also high because this population shows an important variation in EI index, suggesting that some individuals were digging while some other were not. Therefore, although our inferences are somewhat uncertain for this population, the lifestyle of A. mosbachensis may be considered as intermediate between that of extant ones, or as uncertain. It is also possible that this species may have had a lifestyle with no extant analogue (at least when one considers the population we have sampled). Whether or not fossil populations are included, the estimation of character state for lifestyle for the common ancestor of recent clades is consistent among methods: the ancestor of A. amphibius terrestris + populations of A. scherman ssp was intermediate between semi-aquatic and fossorial, while the ancestor of A. amphibius italicus and A. amphibius musignani was semi-aquatic. However the ancestral character reconstruction for deeper nodes is more uncertain: the value for the most recent common ancestor of all A. amphibius and A. scherman was intermediate between semi-aquatic and fossorial population, but in average closer to the “semi-aquatic pole”. Including or not fossils did not change significantly this value. The value of the most common ancestor of all living species or population of Arvicola is more uncertain, but seems to be closer to semi-aquatic populations, meaning that fossorial adaptation are secondary in A. scherman populations. Inclusion of extinct forms did not help to reduce uncertainties, but suggested that this ancestor may have had a lifestyle intermediate between semi-aquatic and fossorial. This, however, is speculative and one would need more fossil populations and living analogues to test our hypotheses. Moreover, uncertain position of fossils and uncertainties in topologies and branch lengths still remain in the phylogeny. Including the sister groups of Arvicola (Microtus, Myodes, and Ondatra) where both lifestyles are also known to occur, could probably shed some more lights concerning the evolution of a semi-aquatic or fossorial lifestyle and its impact on enamel histology of rodents.
Acknowledgements
The authors would like to express their gratitude to Dr. Lutz Maul and the Wiesbaden Museum for the loan of A. mosbachensis specimens. Specimens preparation benefited from the precious suggestions and help from Hayat Lamrous, and Louise Zylberberg Université Paris VI. Norwegian material was kindly provided by Claudia Melis (Centre for Conservation Biology, Norwegian University of Science and Technology, Biology Department, Trondheim, Norway). SEM analyses have been carried out thanks to Dr. Sergio Lo Mastro, Università Roma Tre. Paolo Piras contribution was granted by 2008 post-doc fellow at Pierre et Marie Curie University, Paris (UMR CNRS 7179).
AguilarJ.P.PélissiéT.SigéB.MichauxJ.Occurrence of the Stripe Field Mouse lineage (Apodemus agrarius Pallas 1771; Rodentia; Mammalia) in the Late Pleistocene of southwestern FranceC. R. Palevol.72008217–225AiroldiJ.P.AttrocchiR.MeylanA.Le comportement fouisseur du campagnol terrestre, Arvicola terrestris scherman Shaw (Mammalia Rodentia)Rev. Suisse. Zool.831976282–286AlhaiqueF.BiettiA.Capasso BarbatoL.GrimaldiS.KotsakisT.KuhnS.LemoriniC.ManziG.RecchiA.SchwarczH.P.StinerM.Settlement patterns and susbsistence strategies in the Late Musterian site of Grotta Breuil (M. Circeo Italy)Proc. XIII Intern. Congr. Prehist. Protohist. Sci. Forlì.611998221–228BaudoinC.Le campagnol amphibie. Arvicola sapidusFayardA.Atlas des Mammifères sauvages de France1984SFEPMParispp. 162–163BernardJ.A propos d’un croisement entre Arvicola terrestris terrestris L. et A. terrestris scherman ShawMammalia251961120–121BrinkF.H.A.vanden.A field guide to the mammals of Britain and Europe1967CollinsLondon221 p.Buzas-StephensP.DalquestW.W.Enamel ultrastructure of incisor, premolars and molars in Thomomys, Cratogeomys, and Geomys (Rodentia, Geomyidae)Texas J. Sci.43199165–74CagninM.Genere Arvicola Lacépède, 1799AmoriG.ContoliL.NappiA.Fauna d’Italia. Mammalia II Erinaceomorpha, Soricomorpha, Rodentia e Lagomorpha.2008CalderiniBolognapp. 444–464Centeno-CuadrosA.DelibesM.GodoyJ.A.Dating the divergence between Southern and European water voles using molecular coalescent-based methodsJ. Zool.2792009404–409ChalineJ.Brunet-LecomteP.CampyM.The last glacial-interglacial record of rodent remains from the Gigny karst sequence in the French Jura used for paleoclimatic and paleoecological reconstructionsPalaeogeogr. Palaeoclimatol. Palaeoecol.1171995229–252CorbetG.B.The mammals of the Palearctic region: a taxonomic review1978Cornell University PressLondon and Ithaca314 p.CuboJ.VenturaJ.CasinosA.A heterochronic interpretation of the origin of digging adaptations in the northern water vole, Arvicola terrestris (Rodentia: Arvicolidae)Biol. J. Linn. Soc.872006381–391Cuenca-BescósG.CanudoJ.I.LaplanaC.La séquence de rongeurs (Mammalia) des sites du Pléistocène inférieur et moyen d’Atapuerca (Burgos Espagne)L’Anthropologie1042001115–130Harmon, L., Weir, J., Brock, C., Glor, R., Challenger, W., Hunt, G., 2008. GEIGER. Investigating evolutionary radiations. Bioinformatics 24, 129-131.KoenigswaldW.V.Schmelzstruktur und Morphologie in den Molaren der Arvicolidae (Rodentia)Abh. Senckenberg. Naturforsch. Ges.53919801–129KoenigswaldW.V.Enamel microstructure of rodent molars, classification, and parallelisms, with a note on the systematic affiliation of the enigmatic Eocene rodent ProtoptychusJ. Mamm. Evol.112004127–142KotsakisT.Late Pleistocene fossil microvertebrates of Grotta Breuil (Monte Circeo Central Italy)Quatern. Nova.11991325–332KratochvílJ.Variability of some criteria in Arvicola terrestris (Arvicolidae Rodentia)Acta Sci. Nat. Brno1719831–40LavilleP.E.Étude cinématique du fouissage chez Arvicola terrestris scherman (Rodentia Arvicolidae)Mammalia531989177–189LavilleP.E.CasinosA.GascJ.P.RenousS.BouJ.Les mécanismes du fouissage chez Arvicola terrestris et Spalax ehrenbergi : étude fonctionelle et évolutiveAnat. Anz.1691989131–144Marquet, J. C., 1989. Paléoenvironnement et chronologie des sites du domaine atlantique français d’âge Pléistocène moyen et supérieur d’après l’étude des rongeurs. Thèse, Université de Bourgogne, Dijon, 637 p.MartinT.Incisor Schmelzmuster diversity in South America's oldest rodent fauna and early caviomorph historyJ. Mamm. Evol.122005405–417MasiniF.MaulL.C.AbbazziL.PetrusoD.Independent Arvicola lineages in Italy and Central Europe?MacholanM.BryjaJ.ZimaJ.Abstr. Fourth European Congr.2003MammalogyBrno164 p.MaulL.C.MasiniF.AbbazziL.TurnerA.The use of different morphometric data for absolute age calibration of some south and middle European arvicolid populationPalaeontogr. Ital.851998111–151MaulL.C.RekovetsL.HeinrichW.D.KellerT.StorchG.Arvicola mosbachensis (Schmidtgen 1911) of Mosbach 2: a basic sample for the early evolution of the genus and a reference for further biostratigraphical studiesSenckenb. Lethaea.802000129–147Müller-Böhme, H., 1936. Beitrage zur Anatomie, Morphologie und Biologie der Grossen Wühlmaus. (Arvicola terrestris L. Arvicola terrestris scherman Shaw). Gleichzeitig ein Versuch zur Lösung ihrer Rassenfrage. Arb. Biol. Anst. Land.-u. Forstw. Berl 21, 363–463MusserG.G.CarletonM.D.Superfamily Muroidea3rd Ed.WilsonD.E.ReederD.A.M.Mammal Species of the World,vol. 22005The Johns Hopkins University PressBaltimorepp. 894–1531PanteleyevP.A.The water vole: mode of the species2001NaukaMoscow527 p. [in Russian]ParadisE.ClaudeJ.StrimmerK.APE: analyses of phylogenetics and evolution in R languageBioinformatics202004289–290PiertneyS.B.StewartW.A.LambinX.TelferS.AarsJ.DallasJ.F.Phylogeographic structure and postglacial evolutionary history of water voles (Arvicola terrestris) in the United KingdomMol. Ecol.1420051435–1444ReichsteinH.Beitrag zur systematischen Gliederung des Genus Arvicola Lacépède, 1799Z. Zool. Syst. Evolutionsforsch.11963155–204ReichsteinH.Gattung Arvicola Lacépède, 1799NiethammerJ.KrappF.Handbuch der Säugetiere Europas, II.1982Akademische VerlagsgesellschaftWiesbadenpp. 209–252RensbergerM.KoenigswaldW.V.Functional and phylogenetic interpretation of enamel microstructure in rhinocerosesPaleobiology61984477–495SalaB.Gli animali del giacimento di Isernia La PinetaPerettoC.I reperti paleontologici del giacimento paleolitico di Isernia La Pineta, l’Uomo e l’ambiente. Istituto Regionale per gli Studi Storici del Molise “V. Cuoco”1996Cosmo Iannone EditoreIserniapp. 87–186SaucyF.Arvicola sapidus Miller, 1908Mitchell-JonesA.J.AmoriG.BogdanowiczW.KryštufekB.ReijndersP.J.H.SpitzenbergerF.StubbeM.ThissenJ.B.M.VohralíkV.ZimaJ.The atlas of European mammals2202002T & AD Poyser Ltd Natural HistoryLondon221 pSignorP.W.IIILippsJ.H.Sampling bias, gradual extinction patterns, and catastrophes in the fossil recordSilverL.T.SchultzP.H.Geological implications of impacts of large asteroids and comets on the Earth.1901982Geological Society of America Special Publicationpp. 291–6SteinB.Morphology of subterranean rodentsLaceyE.PattonJ.CameronG.Life Underground2000University of Chicago PressChicagopp. 19–61TaberletP.F.L.Wust-SaucyA.G.CossonJ.F.Comparative phylogeography and postglacial colonization routes in EuropeMol. Ecol.71998453–464TomesJ.On the structure of the dental tissues of the order Rodentia. Philos. Trans. R. Soc. LondonA Math. Phys. Sci.1401850529–567van DamJ.The small mammals from the Upper Miocene of the Teruel–Alfambra region (Spain): paleobiology and paleoclimatic reconstructionsGeol. Utraiectina.15619971–204VenturaJ.Arvicola sapidus Miller, 1908PalomoL.J.GisbertJ.BlancoJ.C.Atlas y libro rojo de los mamíferos de España2007Dirección General para la Biodiversidad-SECEM-SECEMUMadridpp. 405–407VieytesE.C.MorganC.C.VerziD.H.Adaptive diversity of incisor enamel microstructure in South American burrowing rodents (family Ctenomyidae, Caviomorpha)J. Anat.2112007296–302ZagorodnyukI.V.Nomenclature and system of genus ArvicolaPantelejevP.A.The Water Vole. Mode of the Species2001NaukaMoscowpp. 174–92 [in Russian]
Location map of the populations of Arvicola used in the text. 1. A. scherman cantabriae. 2. A. sapidus. 3. A. scherman monticola. 4. A. scherman exitus. 5. A. scherman scherman. 6. A. amphibius italicus. 7. A. amphibius musignani. 8. A. amphibius terrestris. 9. A. mosbachensis extinct (Mosbach, Germany). 10. A. scherman extinct (Bouziès, France). 11. A. amphibius extinct (Grotta Breuil, Italy).
Carte indiquant l’origine des populations citées dans le texte. 1. A. scherman cantabriae ; 2. A. sapidus ; 3. A. scherman monticola ; 4. A. scherman exitus ; 5. A. scherman scherman ; 6. A. amphibius italicus ; 7. A. amphibius musignani ; 8. A. amphibius terrestris ; 9. A. mosbachensis extinct (Mosbach, Allemagne) ; 10. A. scherman extinct (Bouziès, France) ; 11. A. amphibius extinct (Grotta Breuil, Italy).
Phylogenetic relationships among the taxa analyzed in this study based on molecular data (Piertney et al., 2005 and Taberlet et al., 1998). Branch lengths were assigned using the paleontological literature (Cuenca-Bescós et al., 2001, Masini et al., 2003 and Maul et al., 1998) and are proportional to geological time, in m.y. The two-presented phylogenies differ in the placement of A. mosbachensis (on the left: phylogenetic scenario 1, on the right: phylogenetic scenario 2). Nodes labels are used in Table 2.
Relations phylogénétiques entre les taxons analysés dans cette étude basée sur des données moléculaires (Piertney et al., 2005 ; Taberlet et al., 1998). Les longueurs des branches ont été calculées utilisant la littérature paléontologique (Cuenca-Bescós et al., 2001 ; Masini et al., 2003 ; Maul et al., 1998) et sont proportionnelles au temps géologique en millions d’années. Les deux phylogénies présentent deux scénarios différents concernant la position de A. mosbachensis (à gauche le scénario phylogénétique 1 ; à droite, le scénario phylogénétique 2). La numérotation des noeuds est reprise dans le Tableau 2.
SEM photographs of enamel microstructure of extant Arvicola lower incisors (cross sections in A and B, outer tooth surface at right, enamel–dentine junction at left; in C and D: outer tooth surface at left, enamel-dentine junction at right). A: A. scherman cantabriae; B: A. amphibius musignani; C: A. scherman scherman; D: A. sapidus. Scale bar represents 50 μm.
Photographies de sections transversales d’incisives inférieures d’Arvicola actuels (surface externe de la dent à droite, jonction émail - dentine à gauche dans A et B; surface externe de la dent à gauche, jonction émail - dentine à droite dans C et D) obtenues par microscopie électronique à balayage. A : A. scherman cantabriae ; B : A. amphibius musignani ; C : A. scherman scherman ; D : A. sapidus. L’échelle représente 50 μm.
SEM photographs of enamel microstructure of fossil Arvicola lower incisors (cross sections, outer tooth surface at left, enamel–dentine junction at right). A: Arvicola scherman extinct (Bouziès, France); B: A. mosbachensis extinct (Mosbach, Germany). Scale bar represents 50 μm.
Photographies de sections transversales d’incisives inférieures d’Arvicola fossiles (surface externe de la dent à gauche, jonction émail – dentine à droite) obtenues par microscopie électronique à balayage. A : Arvicola scherman extinct (Bouziès, France) ; B : A. mosbachensis extinct (Mosbach, Germany). L’échelle représente 50 μm.
Sampling size (n), mean and standard deviation (sd) for EI values, and lifestyle for living populations (1: fossorial, 0: semi-aquatic).
Taille de l’échantillon (n), moyenne et écart-type (sd) pour les valeurs du EI, et mode de vie des populations actuelles (1 : fouisseur, 0 : semi-aquatique).
Ancestral character reconstructions using different phylogenetic scenarios and estimates of A. mosbachensis lifestyle. AC1: ancestral character estimation of lifestyle using living taxa only. AC2: ancestral character estimation using both living and fossil taxa according to the first phylogenetic scenario and the logistic model for inferring A. mosbachensis lifestyle. AC3: Same as AC2, but using phylogenetic scenario 2. AC4: same as AC2 but using a linear interpolation for A. mosbachensis lifestyle. AC5: same as AC4 but using phylogenetic scenario 2. 0 corresponds to an exclusive aquatic lifestyle, while 1 corresponds to an exclusive fossorial lifestyle. 95% confidence half intervals are provided.
Reconstruction des modes de vie ancestraux utilisant différents scénarios phylogénétiques et estimations du mode de vie de A. mosbachensis. AC1 : Reconstruction des modes de vie ancestraux utilisant seulement des taxons actuels. AC2 : Reconstruction des modes de vie ancestraux utilisant des taxons actuels et fossiles et assumant le scénario phylogénétique 1 et le modèle logistique pour inférer le mode de vie de A. mosbachensis. AC3 : comme dans AC2, mais utilisant le scénario phylogénétique 2. AC4 : comme dans AC2, mais utilisant une interpolation linéaire pour inférer le mode de vie de A. mosbachensis. AC5 : comme AC4, mais utilisant le scénario phylogénétique 2. 0 correspond à un mode de vie exclusivement aquatique, tandis que 1 correspond à un mode de vie exclusivement fouisseur. Des intervalles de confiance 95 % sont fournis.