Mutzig, located in northeastern France (Fig. 1 and Fig. 2), represents an exceptional site for the understanding of the Middle Palaeolithic in the upper Rhine region. The archaeological site of Mutzig is an ancient rock-shelter (now collapsed) located at 340 m ASL at the foot of the Mutzigberg rock mass composed of sandstone, near the Bruche river. The site was discovered in 1992 during construction work in the garden of a house. The site was covered by sandstone rocks, which would have played a key role in the preservation of the archaeological material, protecting it from erosion. Several excavations were undertaken between 1992 and 1996 by J. Sainty (Sainty, 1992 and Sainty, 1993a), and since 2009 by J. Detrey and T. Hauck (2009–2012) and then H. Koehler (since 2013). Several periods of Neanderthal occupation have been identified and yielded Mousterian industries, small and large faunal remains, some of them displaying burning and butchery evidences. OSL dates yielded an age dating to the end of MIS 5 for UE 7a, 8, 9 (Detrey and Hauck, 2011 and Koehler et al., 2013; Preusser et al., in prep.). ESR-U/Th dating is in progress (J.-J. Bahain, MNHN). The “old collection” of large mammal remains (i.e. Sainty's excavations, called Mutzig 1, 2, 7 and 8; Sainty, 1993a and Sainty, 1993b) are represented by several species: Rangifer tarandus, Bison priscus, small bovid (cf. Rupicapra rupicapra?), Equus sp., Cervus elaphus, Megaloceros giganteus, Capreolus capreolus, Mammuthus primigenius, Canis lupus, Vulpes vulpes, Ursus sp., a small carnivore (Auguste, 2012 and Patou-Mathis, 1993). The large fauna coming from the recent 2010–2014 excavations (in association with the small vertebrates) are a little different from the old corpus: Rangifer tarandus, Bison priscus, small bovid (cf. Rupicapra rupicapra?), Equus sp., Coelodonta antiquitatis, Mammuthus primigenius, Canis lupus, cf. Alopex lagopus, Ursus cf. arctos.

Located on the western coast of the Cotentin (Manche, Fig. 1), to the North of the Surtainville beach, the Rozel ancient rock-shelter was revealed by coastal erosion. The site is part of a dune formation formed by the wind during the early stages of the last glaciation and ‘trapped’ in an old creek opening in a rocky cliff composed of schist crossed by quartz veins (Cliquet et al., 2014, Scuvée and Verague, 1984 and Scuvée and Verague, 1988; Fig. 3). The site was discovered by Y. Roupin in 1967, and was then the subject of excavations in 1969 and 1970 by F. Scuvée (Scuvée and Verague, 1984). New excavations directed by D. Cliquet are ongoing since 2012. Investigations conducted on the sedimentary deposits and on the lithic and bone material suggest that the site was occupied by Neanderthals at least twice (two occupation soils; Van Vliet-Lanoë et al., 2006; Fig. 2). OSL dating on the “Grande dune” levels have yielded ages corresponding to the middle or end of MIS 5: around 107–87 ka according to Folz (2000) and 115–102 ka according to Van Vliet-Lanoë et al. (2006). Le Rozel is a major Middle Palaeolithic site in northern France, due to the exceptional preservation of anthropic structures (hearths), bone material (large and small vertebrates), insect remains, Neanderthal footprint, and the originality of its lithic industries (Cliquet, 2012, Cliquet et al., 2014, Scuvée and Verague, 1984 and Van Vliet-Lanoë et al., 2006). Unfortunately, the coastal location means erosional processes threatening the sites future survival. As for Mutzig, the large mammal fauna from the old excavations of Le Rozel is a little bit different from that of the recent ones. Old excavations have yielded remains of Bos primigenius, Equus sp., Cervus elaphus, Megaloceros giganteus, Odobenus rosmarus, a small carnivore (Auguste in Van Vliet-Lanoë et al., 2006). Large mammals from recent excavations are represented by cf. Palaeoloxodon antiquus, cf. Stephanorhinus hemitoechus, Bos primigenius, Equus sp., Cervus elaphus, Capreolus capreolus, Oryctolagus cuniculus.

A total of 17 samples coming from UE 2, 5, 6 and 7c were studied. Each level was considered separately, and samples lacking stratigraphic context were counted but excluded from the analyses. All the sediment samples were water-sieved at 1 mm mesh. A sieving test was carried out on two samples at a smaller mesh (0.5 mm), but no more material was found in the 0.5–1 mm fraction. One sample corresponds to a bucket of ∼ 8 L of sediments, leading to a total volume of sediments of around 0.136 m³ for 17 samples. A total of 341 identifiable remains of small vertebrates were extracted from these studied samples. The small vertebrate bone concentration at Mutzig is thus around 2507 identifiable remains/m³.

A total of 91 samples from the first occupation soil and from the yellow fine sand immediately above and below this occupation soil were studied. Consequently, all samples were grouped together as a whole, however samples lacking stratigraphic context were counted but excluded from the analyses. All the sediment samples were dry-sieved at 0.5 mm mesh and then sorted to extract more than 700 identifiable remains of small vertebrates. One sample corresponds to a bucket of ∼ 10 L of sediments, leading to a total volume of sediments of around 0.910 m³ for 91 samples. The small vertebrate bone concentration at Le Rozel is thus around 875 identifiable remains/m³.

Species identifications were based on direct comparisons with osteological collections at the Museum national d’Histoire naturelle (MNHN, Paris), and on illustrations, morphological and biometric criteria from literature (e.g. Abbassi et al., 1998, Bailon, 1999, Callou, 1997, Chaline, 1972, Chaline, 1974, Chaline, 1988, Chaline and Mein, 1979, Cuenca-Bescos, 2003, Desclaux, 1992, Escudé et al., 2008a, Escudé et al., 2008b, Hanquet, 2011, Heinrich, 1987, Kalthoff et al., 2007, van Kolfschoten, 1990, van Kolfschoten, 1992, Maul et al., 2000, Nadachowski, 1982, Paunescu, 2001 and Tuffreau et al., 1995). Only small terrestrial vertebrates were considered and although bird and fish remains were counted, they were not considered in this study. Identifications and taphonomic observations were performed under binocular microscope. Pictures were taken using a microscope equipped with a digital camera at the Department of Prehistory of the MNHN and at the Morphometry platform of the MNHN. Rodent teeth were measured from digital pictures using TpsDig2 software. Data on the ecology and distribution of species comes from literature (e.g. Aulagnier et al., 2008, Desclaux, 1992, Hanquet, 2011 and Jeannet, 2010a; http://www.iucnredlist.org/). For palaeoecological reconstructions, we used classical indices such as Climatogram (based on the relative frequencies of species (MNI), each species belonging only to one ecological group), Taxonomic Habitat Index (based on the presence/absence of species and their occurrence in several types of habitats, regardless of their relative proportions, avoiding potential taphonomic biases) and Cenogram (considering both large and small mammals according to their weight, the structure of a mammalian community depending on the type of landscape and climatic conditions).

Sites used as examples: Cagny-l’Épinette, (Saalian; Tuffreau et al., 1995); Tourville-la-Rivière (Saalian; Cordy et al., 2003; Stoetzel, unpublished); Biache-Saint-Vaast (Saalian; Chaline, 1978); Saint-Sauveur (Weichselian; Antoine et al., 1995).

Similarities exist between the assemblages from Mutzig and those from Cagny-l’Épinette (Tuffreau et al., 1995), despite the older age of Cagny-l’Épinette estimated to MIS 9 thanks to chronostratigraphic position, lithic industries, large mammals and geochronology, with date around 296 ka, 318 ka, 289 ka and 291 ka (Bahain et al., 2007). Indeed, Cagny-l’Épinette yielded a “modern” microfauna assemblage lacking typical Middle Pleistocene species. The climate was fresh-temperate (Talpa europaea, Sorex sp., Arvicola terrestris, Microtus oeconomus) and mostly open (dominance of M. arvalis/agrestis, absence of Clethrionomys glareolus and Gliridae). In addition, Arvicola SDQ values are around 100 like at Mutzig. The recent age of microfauna from fluviatil levels of Cagny-l’Épinette probably results from taphonomy, with many Arvicolinae elements coming from burrow contexts and, consequently, more recent than the Palaeolithic settlement.

The microfauna from Tourville-la-Rivière (Cordy et al., 2003; Stoetzel, unpublished), dating to the middle of the MIS 7 (Bahain in Auguste, in press), consists of Talpa europaea, Arvicola cantiana-terrestris, Microtus arvalis/agrestis, M. oeconomus; Arvicola SDQ values are higher than at Mutzig (∼ 114). The landscape was mostly open and humid (grassy meadows with bushes and water ponds) and a fresh-temperate climate (transition? interstadial?) with the occurrence of moles and the absence of typical cold rodents such as Microtus gregalis or Dicrostonyx torquatus.

Biache-Saint-Vaast dates to the Saalian period (around 240 ka; Auguste, 2009 and Bahain et al., 2015) and has also yielded “modern” microfauna, including Crocidura sp., Sorex sp., Microtus arvalis/agrestis, M. oeconomus/malei, M. (T.) subterraneus, Arvicola terrestris, Clethrionomys glareolus (Chaline, 1978). The presence of Soricidae and C. glareolus indicates a more temperate phase (although colder than today) than at Mutzig. But a taphonomic bias probably explains the huge difference between chronostraphy, geochronology, large mammals and dating based on microfauna interpretation at Biache-Saint-Vaast.

At Saint-Sauveur, a slightly younger site dating to the Weichselian period (Antoine et al., 1995), the presence of Talpa europaea, cf. Neomys sp., Arvicola terrestris, Microtus arvalis/agrestis indicates temperate climatic conditions, intermediate between an interglacial and a glacial period (absence of cold species, presence of temperate species but absence of typical forest ones). Arvicola values indicate a Wechselian or Late Saalian age, as it is the case at Mutzig. But at Saint-Sauveur, the stratigraphic context, U/Th dating and large mammals (large wolf, horse similar to Late Pleistocene forms; Auguste, 1995) argue for an age around the Eemian-Weichselian transition.

Significantly, none of these sites, even the earlier ones, have yielded Allocricetus/Cricetulus or Lagurus remains, which were found at Mutzig. This could result from the fact that all these sites appear to have recorded more temperate periods than Mutzig, less favorable to the occurrence of such species characteristic of open-arid steppe environments. Alternatively, the difference in the biogeographical location of these sites, with a higher influence of the Ocean for Le Rozel and the other sites, against a clear continental location for Mutzig, may explain these differences. Indeed, some authors such as Moine (2014) have highlighted a clear separation between western coastal environments (flat and poorly vegetated domain, more humid) and the eastern continental ones (hilly domain with more diversified vegetation, more dry), with a persistence of these two domains throughout the Weichselian period.

All the mentioned sites have recorded the presence of species which do not occur anymore in northwestern Europe today (such as M. oeconomus/malei), and which were not found at Le Rozel. It therefore appears that Le Rozel recorded a more temperate episode than the comparison sites. However, the questions remains regarding the absence of Soricinae and typical forest rodents (C. glareolus, arboreal species of the squirrel and dormice families) at Le Rozel, which could result from its littoral location.

Sites used as examples: Romain-la-Roche (Saalian; Bailon, 2010, Guérin et al., 2010 and Jeannet, 2010b); Gigny (Saalian–Weichselian; Chaline et al., 1995, Fabre et al., 2011 and Navarro et al., 2004).

Guérin et al. (2010) show that large mammals from Romain-la-Roche and ESR-U/Th dating give an age around 150–165 ka BP (MIS 6) for levels VI–VII and VIII, however, micromammals indicate a more recent age, around the MIS 5 - MIS 4 boundary (Jeannet, 2010b). Taphonomic analysis of the micromammal assemblage from Romain-la-Roche was not carried out, meaning that faunal mixing cannot be excluded as an explanation for these differences. According to Jeannet (2010b) this site has yielded a high small vertebrate diversity, indicating a cooling of conditions towards fully pleniglacial conditions from the basis to the top of the sequence, very different to that observed at Mutzig and Le Rozel.

The site of Gigny covers a large time period, from the end of the Middle Pleistocene to the Late Pleistocene (∼ 145–14 ka BP), and has recorded several palaeoclimatic changes (Chaline et al., 1995, Fabre et al., 2011 and Navarro et al., 2004). The most suitable levels for comparisons with Mutzig and Le Rozel are levels XXII (pre-Eemian) and XX–XVI (post-Eemian). Level XXII is characterized by a transitional climate, from warmer to more temperate conditions, with the presence of Arvicola terrestris, Microtus oeconomus/malei, M. multiplex, M. agrestis, M. gregalis, Clethrionomys glareolus, Apodemus sylvaticus, Eliomys quercinus, Muscardinus avellanarius, Spermophilus superciliosus. Levels XX to XVI–XV have recorded a progressive cooling, with the arrival of Lagurus lagurus, Dicrostonyx torquatus and Marmota (sparse occurrence), Cricetidae, Dipodidae, an increase of M. gregalis and a decrease of temperate forest taxa (Gliridae, Apodemus). The micromammal assemblage and Arvicola measurements of Mutzig are close to those of levels XX–XIXb of Gigny (A. terrestris, M. oeconomus/malei, M. arvalis/agrestis, M. gregalis, L. lagurus), however the presence of M. multiplex was recorded at Gigny and not at Mutzig. In addition, several Gigny levels have yielded M. arvalis/agrestis, A. terrestris and Apodemus remains as Le Rozel, but always associated with several other species not recorded at Le Rozel.

Sites used as examples: Norton Farm (England, Saalian; Bates et al., 2000); Lower Thames (England, Saalian–Weichselian; Sutcliffe and Kowalski, 1976); Plaidter Hummerich A (Germany, Saalian; Kalthoff et al., 2007); Wannenköpfe-West (Germany, Weichselian; Kalthoff et al., 2007); Burgtonna (Germany, Weichselian; van Kolfschoten, 2000); Steinheim an der Mur (Germany, Weichselian; van Kolfschoten, 2000); Stuttgart-Untertürkheim (Germany, Weichselian; van Kolfschoten, 2000); Maastricht-Belvédère (Netherlands, Saalian–Weichselian; van Kolfschoten, 1985 and van Kolfschoten, 1993).

Norton Farm, England (Bates et al., 2000), yielded only remains of Microtus oeconomus. However, their association with pollens, large mammals and molluscs indicate a cold-stage of the late MIS 7 or early MIS 6, too old to be correlated with Mutzig or Le Rozel. Few micromammal assemblages in the UK date to MIS 6 or early MIS 5. Most assemblages come from the Thames terraces (few from stratified cave deposits), which are of interglacial or interstadial age, because most cold-stage deposits are now submerged (Sutcliffe and Kowalski, 1976).

The German sites of Plaidter Hummerich A (Late Saalian) and Wannenköpfe-West (Eemian–Weichselian boundary) have both yielded microfauna indicating a temperate transitional climate, and an open meadow landscape with wet areas and some wooded areas (Kalthoff et al., 2007). Le Rozel shows a high degree of similarity with Wannenköpfe-West: dominance of Microtus arvalis/agrestis and Arvicola terrestris (SDQ ∼ 98.3, m1 length ∼ 3.87 mm), presence of Talpa europaea, Apodemus and several amphibians and squamates. However, Wannenköpfe-West has also yielded Soricinae, Dipodidae, M. multiplex and M. oeconomus, indicating rather open grassland landscape with meadows along water basins and with restricted wooded areas. The authors raise an interesting point: “Thus, the studied fauna shows extremely few forms of both cold–dry and humid–warm conditions, and most taxa are generalized in their ecological preferences. This fact together with the predominance of few arvicoline species suggests transitional climatic conditions either from cold to warm or from warm to cold” (Kalthoff et al., 2007, p. 620). These data contrast with those from older Middle Pleistocene sites (such as Tönnchesberg I, Schweinskopf-Karmelenberg, Wannen) whose microfaunas are characteristic of open, cold and arid steppes (Spermophilus, Cricetulus migratorius, Dicrostonyx gulielmi, Lagurus lagurus, Microtus gregalis) (Kalthoff et al., 2007). Mutzig is closer to the latter sites, with the presence of Spermophilus sp., Allocricetus/Cricetulus sp., L. lagurus and M. gregalis. However, Mutzig lacks typical cold species such as “true” lemmings.

Saalian levels of Maastricht-Belvédère (3 and 4) in the Netherlands correspond to a more temperate phase than Mutzig (with notably the presence of Erinaceus davidi, Talpa europaea, several Crocidurinae and Soricinae species, Eliomys quercinus, Clethrionomys glareolus, Arvicola cantiana/terrestris, Microtus arvalis/agrestis, Apodemus sylvaticus, A. maastrichtiensis…) (van Kolfschoten, 1985 and van Kolfschoten, 1993). However, Maastricht-Belvédère 5 presents a colder fauna, that includes Spermophilus cf. undulatus, Cricetulus migratorius, Dicrostonyx torquatus, Microtus oeconomus, M. gregalis, but with the persistence of more temperate taxa such as Talpa europaea and Arvicola cantiana/terrestris (van Kolfschoten, 1985), which is very similar to that observed at Mutzig, despite the fact that at MB5, Arvicola are slightly more evolved and cold taxa are more represented.

van Kolfschoten (2000) cites several German sites documenting the Eemian-Weichselian transition. The Eemian age of most of the deposits is mainly based on the interglacial nature of flora and fauna (including the presence of Soricidae, Gliridae, woodmice, bank voles) and the evolutionary stage of Arvicola. Palaeontological data indicate that the environment was a mosaic landscape with both woodland and open areas, comparatively palaeobotanical data indicate an extensive forest cover. Early Weichselian faunas from Bugtonna, Stuttgart-Untertürkheim or Steinheim an der Murr are markedly different, with a higher frequency of species inhabiting cold steppes (Ochotona pusilla, Spermophilus citelloides, Lagurus lagurus) presenting thus similarities with Mutzig.

As observed for the sites in northern and northeastern France, faunas from Germany and the Netherlands that date to the early Late Pleistocene appear “modern” with no representation of Allocricetus or other Middle Pleistocene relictual species. However, Cricetulus migratorius appears sporadically.

The microfauna from the Saalian sites considered here (such as Cagny-l’Épinette, Tourville-la-Rivière, Biache-Saint-Vaast, Plaidter Hummerich A, Maastricht-Belvédère 3-4, Gigny XXII) but also from Early Weichselian sites (such as Saint-Sauveur, Wannenköpfe-West) are all very similar and are composed of “modern” species, characteristic of a transitional climate, relatively temperate but cooler than today. Mania and Mania (2008) and Kalthoff et al. (2007) have previously noted that vertebrate faunas from the Late Saalian and the early Weichselian can be defined as “transitional faunas” including at the same time cold, ubiquist, thermophilous and steppic species. This is also what we observe at Mutzig and Le Rozel. Often the evolutionary stage of Arvicola allows the refinement of an age of an assemblage, but biometric values are generally similar for the end of the Saalian period and the beginning of the Weichselian period (van Kolfschoten, 1992). These observations do not help to refine the age attribution of our sites, especially Mutzig, which has the same “common background” as most of the other sites (Talpa europaea, Arvicola terrestris, Microtus arvalis/agrestis, M. oeconomus) and often the same biometric values for Arvicola. Mutzig also yielded other species, more characteristic of cold steppes (Ochotona, Spermophilus M. gregalis, Allocricetus/Cricetulus, L. lagurus). This type of assemblages is found at Maastricht-Belvédère 5 and Gigny XX–XIX (Weichselian) but with some differences in the evolutionary stage or the presence/absence of certain species. Thus Mutzig would correspond to a relatively cold-arid period, but not fully pleniglacial, of the Late Saalian or Early Weichselian, a period that is poorly documented in northwestern Europe concerning microvertebrates. Le Rozel would correspond to a more recent phase of the Early Weichselian, and to a more temperate period, but not fully interglacial, with microfauna similar to that of Wannenköpfe-West for example.