Reappearance of Galerix (Erinaceomorpha, Mammalia) at the Middle to Late Miocene transition in South Germany: biostratigraphic and palaeoecologic implications

The presence of Galerix molars in the South German fossil locality Hammerschmiede 3 is interpreted as evidence for a re- immigration of West European origin into the North Alpine Foreland Basin at the transition of the Middle to Late Miocene. The brief re-appearence of Galerix in southern Germany can be used as a biostratigraphic marker that allows promising correla- tions between local biostratigraphic subdivisions from Spain and Germany, suggesting that, contrary to previous thought, the Hammerschmiede locality may antedate the hipparionine hors-es’ appearance event. Based on the supposed climatic adaptation of galericine taxa and lower vertebrate record, it is hypothesized that short term climatic fluctuations occurred in South Germany around the time of the first appearance of the hipparionine horses in Europe.


Introduction
The Middle Miocene is characterized by major global climate changes, including the end of the Miocene climatic optimum (Böhme, 2003) due to global cooling (Shevenell et al., 2004). In Europe, the end of the Middle Miocene is subjected to an aridity peak, followed by warm conditions and considerably more precipitation than present during the Tortonian ('washhouse' climatic phases, Böhme et al., 2008). Coincident with these climatic events, a major reorganization in the European mammal faunas takes place (Casanovas-Vilar et al., 2010). For instance, a new wave of hominoids entered Europe (Casanovas-Vilar et al., 2008a).
From a biostratigraphical perspective, the most important immigrants at the boundary between Middle and Late Miocene are the hipparionine horses. The transition from Middle to Late Miocene is well documented in Spain, in sections with both large and small mammals in the Daroca-Calamocha area of the Teruel Basin (Álvarez , and the Vallés-Penedés Basin (Casanovas-Vilar et al., 2006;Casanovas-Vilar, 2007). Unfortunately, the entrance of these perissodactyls is not accompanied with major changes in the small mammal faunas (Casanovas-Vilar et al., 2010).
In sharp contrast to the Spanish record, the transition from Middle to Late Miocene is poorly documented in Germany and in Central Europe in general. This hampers correlations through Europe and thus the understanding of large scale patterns in faunal distribution. The German locality Hammerschmiede is one of the few small mammal localities of North Alpine Foreland Basin (NAFB) that represents this time interval. Traditionally, it is placed in MN 9 (earliest Late Miocene), but as no Hipparion s.l. was found in the locality, there is no direct evidence for this designation. In the NAFB, the period is marked by the end of the molasse sedimentation (Frisch et al., 1999), the uplift of the southern part of the Bohemian massive (Ziegler and Dèzes, 2007), the uplift of the eastern Alps (Kuhleman, 2007) and change in the fluvial direction in the basin from originally east-west to west-east. Hammerschmiede sediments (Fig. 1, 6 km NNW of Kaufbeuren, Germany, see Mayr and Fahlbusch (1975) for details) derive from to the youngest deposits of the Upper Freshwater Molasse (UFM) of the NAFB. These deposits are referred to as the Obere Serie (Dopp ler, 1989;Doppler et al., 2005), which is biostratigraphically defined by the presence of the largesized Deinotherium giganteum Kaup, 1829 (Dehm, 1955). The UFM deposits of Hammerschmiede are known for their rich fossil content, dominated by small mammals, ectothermic tetrapods and fishes (fossil layer 1, Mayr and Fahlbusch, 1975;Fahlbusch and Mayr, 1975;Prieto and Rummel, 2009b). The layer of Hammerschmiede 3 corresponds to an erosional channel rich in lignitic pebbles near the top of the sequence.
In the absence of a link with the entrance of the hipparionine horses, mammal palaeontologists had to come up with another marker for MN 9 (Mammal Neogene, De Bruijn et al., 1992;Mein, 1999), the first mammal unit of the Late Miocene. The presence of the highly specialized cricetid rodent Microtocricetus molassicus Mayr, 1975 in Hammerschmiede (Fahlbusch andMayr and Fahlbusch, 1975) is the usual argument to place the locality in MN 9 as the species is traditionally considered as to be typical for this faunal unit in Central and Eastern Europe (De Bruijn et al., 1992: 70;Mein, 1999).
The discovery of Galerix molars in Hammerschmiede 3 is remarkable. The genus was believed to have become extirpated in the region halfway the Middle Miocene (Ziegler, 2006a), while surviving up to the Late Miocene in Spain ( Van den Hoek Ostende, 2001;Van den Hoek Ostende and Furió, 2005). Given the good record of the latter part of the Middle Miocene in Germany, we have to assume that Galerix remigrated into the region. In this paper we explore the stratigraphical and environmental conclusions that can be drawn from this unexpected re-occurrence of the gymnure.

Material and methods
The descriptive terminology for Erinaceidae molars follows Engesser (1980). The orientation of the elements for measuring is after Prieto et al. (2010a).
The separation of the mammal 'zones' MN 7 and MN 8 follows the concept of .

A short overview of the Galericini of Central and Western Europe
Before discussing the Galerix fossils from Hammerschmiede and their biogeographical and stratigraphical consequences, we first present the current state of knowledge of the Galericini record in Central Europe and Spain. Before we can go into that, however, we need to note that there still is an important taxonomical problem that needs to be resolved.

Taxonomic comments on the genera Parasorex and Schizogalerix
The evolutionary history of Schizogalerix, as well as the systematic attribution of the oldest species either to Schizogalerix, Parasorex or Galerix, are still under debate (Van den Hoek Ostende, 2001;Ziegler, 2003Ziegler, , 2005Doukas and Van den Hoek Ostende, 2006). In contemporary taxonomic interpretations, Schizogalerix is not recorded in the late Middle/earliest Late Miocene (MN 7/8, MN 9) deposits from Western Europe (Spain, France). It is, however, important to note that the Spanish Parasorex material shares some similarities with Schizogalerix, as recorded in Central Europe (Engesser, 1980;, especially the configuration of the mesostyle of the M1 and M2 (cf. De Jong, 1988: pl. 1, figs 4, 6;Engesser, 1980: fig. 4b). This suggests possible differences in the taxonomical interpretation between western European and central/eastern European scholars, at least concerning the species Parasorex socialis and S. voesendorfensis. In the following, if not specified, P. socialis will be understood as proposed in Van den Hoek Ostende (2001).
We include the specimens from Jamm (Austria), originally described as Galerix exilis (Thenius, 1949) and later interpreted as Parasorex socialis (Ziegler and Daxner-Höck, 2005;Ziegler, 2006a), into Schizo-galerix, based on the strong posterior cingulum that directly joins the entoconid on the m1 and m2, and the morphology of the mesostyle. Rzebik-Kowalska and Lungu (2009) record the presence of Parasorex socialis in the Moldavian MN 10 locality of Kejnar. No M1 or M2 was found, so the identification was based on the absence of a connection between the posterior cingulum and the entoconid on m1 and m2. However, their illustrations (o.c., figs 4e-f) show that this connection was indeed present, thus refuting the identification as P. socialis. The Kejnar assemblage is very interesting, as its galericine is more primitive than the Schizogalerix of Buzor 1, a locality that the authors correlate to MN 9.
Although Parasorex socialis is reported in literature from the upper part of the Middle Miocene in Romania (Feru et al., 1979(Feru et al., , 1980 and Serbia (Markovic, 2003), the specimens are not described in detail, so their identity cannot be checked. Thus, the presence of P. socialis east from the NAFB has not been convincingly demonstrated.
Regardless of the generic attribution of the Spanish Parasorex material, the immigration of Schizogalerix into the NAFB must have come from the eastern part of Europe/Anatolia, as is evident from the good fossil record in the larger part of Middle Miocene from the Turkey (Engesser, 1980;Selänne, 2003). In the Middle Miocene, Schizogalerix anatolica is known from Mátraszölös (Hungary, Gál et al., 1999). Some specimens are also recorded in the Middle Miocene from Greece (Doukas, 2005), MN 9 from Slovakia (Sabol et al., 2004, and see references in Fejfar and Sabol, 2005), and MN 9 from Moldova (Rzebik-Kowalska and Lungu, 2009). The genus is also present in the early Sarmatian of Rumania in Tăşad (Mészáros, written communication in Hír et al., 2001).

Galerix, Parasorex and Schizogalerix: current state of knowledge in Germany
The first representatives of Galericini are recorded in Germany in the MN 3 fissure fillings of the Swabian and Franconian Alb and in Frankfurt/Nordbasin (Zieg ler et al., 2005). Most of this material is identified as Galerix aurelianensis, but some smaller specimens have been questionably attributed to G. symeonidisi. In MN 4 fissure fillings, G. aurelianensis certainly coexists with G. symeonidisi, whereas in contemporaneous stratified localities of the NAFB only the latter species is found (e.g., locality Raucheröd, Ziegler and Fahlbusch, 1986). In MN 5 localities, all from fluviatil deposits, one Galericini species is recorded, which is incorporated by Ziegler and Fahlbusch (1986) into the G. symeonidisi-G. exilis lineage. Van den Hoek Ostende and Doukas (2003) do not recognize this lineage in Spain and proposed that G. exilis gradually replaced the G. symeonidisi (comments in Ziegler, 2005: 138). Scarce evidence of G. aurelianensis-like forms is recorded from the late Karpatian of Sandelzhausen (Zieg ler, 2000) and possibly Maßendorf (Galerix sp. in Schötz, 1988). In MN 6 localities (stratified and fissures; e.g. locality Goldberg, Ziegler 1983) G. exilis is mostly the only Galericini. A larger Galerix species was recently described in the fissure filling Petersbuch 68 (probably latest MN 6, Prieto and Rummel, 2009a).
At the base of the MN 9, Schizogalerix voesendorfensis (Rabeder, 1973) co-occurs with Schizogalerix sp. in the Swiss locality Nebelbergweg .  and Ziegler (2006a) listed P. socialis in the MN 9 faunas from Aumeister and Grosslappen near Munich. Much of the original material from these localities, published by Stromer (1940), was lost during the World War II. On the basis of the remaining fossils stored in the BSPG, we find no evidence for the presence of P. socialis in MN 9 localities. The material from Grosslappen is too poorly preserved (BSPG 1927VIII 20, 1933 to be confidently identified. According to us, the well-preserved mandible from Aumeister belongs to the genus Schizogalerix, as is indicated by the strong posterior cingulum that directly joins the entoconid on the m1 and m2. The sizes of the teeth (p3: L: 1.86 × W: 0.98; p4: L: 2.03 × W: 1.55; m1: L: 3.00 × W1: 1.79 × W2: 2.01; m2: L: 2.30 × W1: 1.79 × W2: 1.84) agree with those of Schizogalerix sp. from Nebelbergweg .

Galerix exilis and Parasorex socialis in Spain
The first occurrence of Parasorex socialis in Spain is in the Vallès-Penedès Basin in San Quirze, a locality that is correlated to the Megacricetodon ibericus -Democricetodon crusafonti Zone (corresponding to the MN 8 sensu Mein andGinsburg 2002, Casanovas-Vilar et al., 2008a). This zone extends from 11.9-11.5 to 10.9 Ma (Casanovas-Vilar et al., 2008b) and may correspond to the upper part of the zone G3 of the Calatayud-Daroca and Teruel Basins (Casanovas-Vilar, 2007). Galerix exilis is not recorded in the Vallès-Penedès. Parasorex socialis is also documented in the Teruel basin with its first occurrence in Nombrevilla 2 (11.5 Ma in Daams et al., 1999;11.69 Ma in Van Dam et al., 2006).
The presence of Galerix exilis in the Daroca-Calata yud area of the Teruel Basin was noted by De Jong (1988), who described material from the late Middle Miocene. The species was present earlier, and, according to Van den Hoek Ostende and Doukas (2003), gradually replaced G. symeonidisi. The older material of G. exilis has not yet been described in detail, and the entire sequence is currently under study. According to Van den Hoek Ostende and Doukas (2003 : Figs 1-2), the species is certainly present from zone D upwards (MN 5). The last occurrence of G. exilis depends on the dating of Carrilanga 1 (10.5 Ma according to Daams et al., 1999, 11.57 Ma according to Van Dam et al., 2006). As the biostratigraphic correlation of Carrilanga proves to be so difficult we place the last occurrence of G. exilis in Spain at 11.5 Ma, based on its occurrence in Nombrevilla 2, pending ongoing research of the Museo Nacional de Ciencias Natureles (Madrid) on the Aragonian/Vallesian sections in the area.
The Hammerschmiede galericine Figure 2 shows a selection of the molars from the Hammerschmiede gymnure. The upper molars, which are clearly wider than long and posses a distinct, crescent-shaped metaconule, identify the Hammerschmiede gymnure as a member of the Galericini. Because most of the M1 (five out of eight) and M2 (one out of two) have a short posterior arm of the metaconule, the P3 have one lingual cusp, and there is no division of the mesostyle in the M1 and M2, the material must be attributed to Galerix (Van den Hoek Ostende, 2001).
Since a substantial fraction of the molars are worn in our sample, the relative abundance of specimens having a protocone-metaconule connection on M1/M2 is difficult to establish. Most of the specimens do not have this ridge complete, but it is often incipiently present. On one considerably damaged M2, the protocone-metaconule connection is present and the hypocone is isolated. The connection of the posterior arm of the metaconule with the posterior cingulum, if present, is weak. The metaconid of the single unworn p4 is about half the height of the protoconid. On the two other specimens, the metaconid seems to be relatively high due to the abrasion of the teeth. The cuspids are worn on their posterior side.
One M1 from Hammerschmiede (Fig. 2D) has a curved anterior arm of the protocone, so that the anterior part of the molar seems to be compressed in comparison with the other specimens. Given the limited number of molars, this tooth is here interpreted as an aberrant specimen.
Compared to Galerix exilis from older localities in southern Germany (Ziegler, 1983), and from the type locality Sansan (Engesser, 2009), the Hammerschmiede 3 assemblage differs in having a long posterior arm of the metaconule on several M1 and M2 (three out of eight M1, one out of two M2). It also differs in the morphology of the P3, as the three preserved premolars from Hammerschmiede 3 have a better-developed lingual part of the tooth. Furthermore, the parastyle and the posterior arm of the para-  Jong, 1988), and are thus assigned to G. cf. exilis. The measurements for the Hammerschmiede Galerix are given in Table 1.

Implication for the Middle/Upper Miocene biostratigraphy in the NAFB
The first occurrence of hipparionine horses in Europe defines the beginning of MN 9 (e.g., Mein, 1999). In the Vallès-Penedès Basin (Catalonia, Spain) the oldest occurrence of Hipparionine horses is in the locality of Creu Conill 22 in the Can Guitar 1 section, and is correlated to the base of the chron C5r.1n, about 11.1 Ma (Garcés et al., 1996(Garcés et al., , 2003Casanovas-Vilar et al., 2006). This is somewhat older than the estimate for the first occurrence of hipparionine horses in the Daroca-Calatayud area in Central Spain (10.7-10.8 Ma in Garcés et al., 2003) and slightly younger than the estimated age of this event in Austria (11.2 Ma in Daxner-Höck, 1996;Rögl and Daxner-Höck, 1996). In the NAFB Hippotherium primigenium is rare, reported only from the large mammals localities Charmoille (Kälin, 1993), Lugnez (Erzinger, 1943), Hausruck (Thenius, 1952) and Höwenegg (e.g., Tobien, 1938Tobien, , 1982. The latter fossil locality is dated 10.3 Ma (Steininger et al., 1996: 14) to 10.8 +/-0.4 Ma (Baranyi et al., 1976;Becker-Platen et al., 1977). Thus, the presence of Galerix cf. exilis challenges the validity of the earliest Late Miocene correlation of Hammerschmiede 3. We propose that Galerix must have entered the NAFB before its extinction in Spain, i.e., before or at least at 11.5 Ma (the age of Nombrevilla 2). The alternative scenario, in which Galerix survived in France in between its extirpation in Spain and entrance into the NAFP is deemed highly unlikely, but cannot be precluded given the current knowledge of the French fossil record.
Following this interpretation, there is no strong argumentation against Hammerschmiede being older than the first appearance of Hippotherium in the NAFB. However, this inference challenges the biostratigraphic value of Microtocricetus and the relative dating of some fossil faunas (see complementary discussion in Prieto et al., 2010b).
Correlating Hammerschmiede to the late Middle Miocene may also have consequences for the correlation of other localities in the NAFB. Based on the evolutionary stage of the cricetid rodent Collimys, Prieto and Rummel (2009b) proposed that Hammerschmiede is older than the Swiss locality of Nebelbergweg, which has also been considered an MN 9 locality . This correlation is mainly based on the presence of Schizogalerix voesendorfensis in the Swiss locality, because until now the oldest occurrence of Schizogalerix voesendorfensis in Austria was in its type locality (Vösendorf/Inzersdorf), dated between 10.3 and 10.6 Ma (Daxner-Höck, 2004;Ziegler and Daxner-Höck, 2005). Recent discoveries, however, indicate that this species is already present in the late Sarmatian s. str. (late Middle Miocene) of the Styrian Basin (around 12.2 Ma, Prieto et al., 2010a). Thus, the locality of Nebelbergweg cannot unambiguously be correlated to the early Late Miocene. Similarly, the German locality Aumeister cannot be confidently correlated based on Schizogalerix. However, a single Collimys m1 was found in Aumeister (Fig. 3). This tooth agrees morphologically with C. longidens from Nebelbergweg and its size (1.86 × 1.21) is intermediate between C. hiri from Hammerschmiede and C. longidens. This suggests that the locality Aumeister is probably slightly younger than Hammerschmiede. Excluding ecological and/or taphonomical bias, it could signify that between the last occurrence of Para sorex socialis (MN Fig. 3. Collimys cf. longidens Kälin andEngesser, 2001 from Aumeister. Left m1 (BSPG 1926 V 21). 8; Engesser, 1972;Kälin, 1993;Prieto, 2007) and the first appearance of Schizogalerix in Nebelbergweg, the faunas of the NAFB could have been characterized by Galerix cf. exilis.

Ecological implications
Assuming that the ecology of the fossil galericines was similar to that of the extant representatives of this subfamily as found in Southeast Asia, they should have been adapted to forest environments close to water (Ziegler, 2005(Ziegler, , 2006b. This certainly seems to be true for Lantanotherium, the only member of the subfamily in the European Miocene that is not included in the tribe Galericini (Furió et al., 2011). Lantanotherium often co-occurs with the presumably wateradapted Dimylidae (Plesiodimylus), e.g. in the German locality of Hambach 6C (Ziegler and Mörs, 2000; mean annual precipitation (MAP)>1100 mm, M. Böhme unpublished data estimated after the method of Böhme et al., 2006) and of Maßendorf (Schötz, 1988;mean MAP: 887mm, M. Böhme unpublished data), in Devínska Nová Ves -Bonanza (Slovakia, Sabol, 2005), in Tobel Hombrechtikon (Switzerland, Bolliger, 1992), in Götzendorf (Austria, Ziegler, 2006b, MAP: 1.303mm (Böhme et al., 2006)) and in Bełchatów B (Poland, Rzebik-Kowalska, 1993. Its distribution is mostly restricted to Central Europe. In Spain, the genus is present in the more humid Vallès-Penedès Basin (Casanovas-Vilar and Agusti, 2007;Furió et al., 2011), but absent in the inner basins (Van den Hoek Ostende and Furió, 2005). All indications are that Lantanotherium preferred humid environments.
In contrast, the Galericini seemed to have been more generalists (Furió et al., 2011), as is already indicated by their almost ubiquitous presence in Miocene faunas. Based on functional morphological observations and faunal associations, Van den Hoek Ostende (2001) proposed that Parasorex, Schizogalerix and probably Galerix symeonidisi were adapted to a more herbivorous diet than the other members of the tribe. The preferred habitat of the various species of Galerix has been debated (Van den Hoek Ostende and Doukas, 2003;Van den Hoek Ostende, 2003;Van den Hoek Ostende and Fejfar, 2006). Certainly, Galerix exilis must have been able to cope with dry environments, such as the semiarid climate of the Middle Miocene German localities of Goldberg and Steinberg (Bolten et al., 1976;Ballmann, 1979;Ziegler, 1983) and, e.g. Somosaguas Sur in Spain (Luis and Hernando, 2000). It is one of the few insectivores which remains present in the sections of the Aragonian type-area in Spain following a dramatic decrease in humidity near the boundary of local zones C and D at approximately 15.9 Ma (Van der Meulen and   fig. 7, with age corrected according to Daams et al., 1999).
Based on comparisons of faunal associations of the late Middle Miocene fissure fillings and stratified fossil localities of Germany, Prieto (2007) proposed that Parasorex socialis was not necessarily dependent on humid or forest-covered habitat. The occurrence of P. socialis in the NAFB in Steinheim, between 14 to 13.5 Ma, is almost contemporaneous with the local disappearance of Galerix exilis. This could suggest different ecological preferences for the two genera. Nevertheless, Galerix and Parasorex can co-occur in a single locality as shown in La Grive (Mein and Ginsburg, 2002;Maridet, 2003), Petersbuch 68 (Prieto and Rummel, 2009a) and Carrilanga (De Jong, 1988).
Notably, the migration of Parasorex in the NAFB coincides with a global cooling event around 14 to 13.5 Ma (Shevenell et al., 2004;Böhme, 2003). This climatic deterioration resulted in a stronger latitudinal temperature gradient in Europe and marks the beginning of a more distinct climatic zonation on the continent. The thermophilic ectotherms survived only in southern regions (south of 37° N palaeolatitude, Böhme, 2003). The cooling event has also been observed in the Teruel Basin (Van der Meulen and   fig. 7, absolute ages corrected according to Daams et al., 1999). As we noted, Galerix exilis survived here during the late part of the Middle Miocene. It becomes extinct near the boundary between local zones G3 and H, i.e., the Aragonian/Vallesian boundary. This time frame coincides with a minor cooling, at a time that humidity increases again (Van der Meulen and . Therefore, we postulate that Parasorex socialis (sensu Engesser, 1972) was a cooler temperature adapted faunal element, whereas Galerix exilis probably preferred higher temperatures.

Consequences for Hammerschmiede
Galerix cf. exilis is a rare element of the fauna from Hammerschmiede 3 and is missing in the two other fossil layers of the locality. Putting aside side the possible effects of taphonomic processes, we can hypothesize: 1] that during the sedimentation of the Hammerschmiede section the climate was somewhat warmer and/or dryer and Galerix cf. exilis represents a new immigrant, which was not well established in the faunas. This scenario is in conflict with the high precipitation estimates from the herpetological community of Hammerschmiede (Böhme et al., 2006). 2] that the environment of Hammerschmiede 3 was not optimal for the Galerix cf. exilis, but the (re-) immigration of Galerix in the NAFB occurred somewhat earlier during an episode in which the environment differed from that found at Hammerschmiede. Such an episode presented unfavourable conditions for Parasorex socialis (sensu Engesser, 1972), which lead to its disappearance.
The second hypothesis implies a brief but distinct temperature increase and/or decrease in precipitation between Anwil and Hammerschmiede. Such decrease in precipitation is recorded in the NAFB. Böhme et al. (2008) demonstrate that the mean annual precipitation in the region decreased significantly between 12.7 and 11.5 Ma (Sarmatian). The rare presence of crocodiles in Anwil (Böhme, 2003) shows that there were indeed episodes with increased temperature during the later part of the Middle Miocene.

Conclusion
Galerix cf. exilis is a short-duration Spanish (re-) immigrant in the NAFB (transient sensu Van der Meulen et al., 2005) and Parasorex socialis a central/ eastern European immigrant into Spain. In this hypothesis Hammerschmiede should be older than the last occurrence of G. exilis in Spain (11. 5 My), and thus should be older than traditionally accepted, probably antedating the Middle to Late Miocene transition. The re-appearance of Galerix in the NAFB suggests climatic fluctuations at the end of the Middle Miocene, leading to temporary warmer and/or dryer conditions. At the actual stage of knowledge 1) we conclude that Schizogalerix appears in the NAFB following the re-entrance of G. exilis, and 2) Microtocricetus can no longer be maintained as marker for MN 9.
Although a revision of the genera Galerix, Parasorex and Schizogalerix is necessary at the European scale, the re-appearance of Galerix cf. exilis in the NAFB is a very promising event in local and largescaled biostratigraphical terms. Indeed, this migration can be correlated with major climatic global events and thus allows us to follow the consecutive faunal interchanges near the limit Middle to Late Miocene.