Cueva Victoria is a karstic cavity located on the eastern slopes of the Cerro de San Ginés (massif of San Ginés de la Jara), near the city of Cartagena (Murcia) and a small lagoon adjacent to the Mediterranean Sea, called Mar Menor (Fig. 1). During the Early Pleistocene, the cavity opened out and filled up with detrital material dominated by heterometric breccias, coarse sands and clays, constituting the so-called fossiliferous breccia. The cavity at that time was used as a hyena's den, as evidenced by the large amount of coprolites and gnawed and broken bones (Gibert et al., 1992).

The age of the fossiliferous layers has been established from the biochronology (see Blain et al. (2008a) for a discussion) and palaeomagnetism (the top of the breccia (8 m) is slightly older than the base of the Jaramillo normal event estimated at 1072 ± 2 ka (Blain et al., 2008a and Gibert et al., 2006).

Until now Cueva Victoria has yielded a large amount of vertebrate fossils that document the richness of the Early Pleistocene biomes of southeastern Spain: more than 80 species have been described (Blain et al., 2008a and Gibert et al., 2006), with at least 39 bird species (Sánchez Marco, 2004). Of great relevance is the occurrence of the primate Theropithecus (Gibert et al., 1995), which suggests a possible dispersal route through the Strait of Gibraltar from North Africa to Europe.

According to the geological work of Montenat (1977) on the palaeogeography of the area, the seashore was closer to the cavity at the end of the Early Pleistocene than today, as shown by the presence of marine mammal fossil remains such as Phoca (Gibert et al., 1992 and Gibert et al., 2006) and four unpublished cetaceous vertebrae.

The amphibian fossil remains used for this study consist of disarticulated elements collected by water-screening the sediments obtained during the palaeontological excavation of the Cueva Victoria cave during the field campaigns of the years 1984–2009. The fossils are provisionally housed in the collections of the Museo Arqueologico Municipal de Cartagena (Spain). The green toad remains recovered from Cueva Victoria correspond for the moment to 498 elements that represent almost all the parts of the skeleton: 1 maxilla, 2 sphenethmoids, 1 parasphenoid, 2 pterygoids, 3 frontoparietal-prootic-exoccipital complexes, 4 mandibles, 56 vertebrae, 5 sacral vertebrae, 19 urostyles, 6 clavicles, 14 scapulae, 4 coracoids, 111 humeri, 59 radioulnae, 78 ilia, 25 femurs, 67 tibiofibulae, 19 tarsals and 20 phalanges. A complete osteological description of all the taxa present in Cueva Victoria is given in another publication (Blain, in press). Note that previous publications on Cueva Victoria amphibians and squamate reptiles were based on the material housed in the collections of the Museu de Geologia de Barcelona, and that no evidence of green toads was found (Blain, 2005, Blain, 2009, Blain and Bailon, 2006 and Blain et al., 2008a).

Comparisons have been drawn mainly using publications (Bailon, 1991, Bailon, 1999, Bailon and Hossini, 1990, Blain, 2005, Böhme, 1977, Delfino et al., 2009, Hodrova, 1986, Hossini, 2001, Hossini, 2002, Ratnikov, 2001 and Sanchíz, 1977, 1998) and using the dry skeleton collections of the Museo Nacional de Ciencias Naturales (Madrid, Spain), of the Muséum national d’Histoire naturelle (Collections d’Anatomie Comparée, Paris, France) as well as Blain's and Bailon's personal collections.

In this work the fossils from Cueva Victoria have been compared with fossil and extant Eurasian and North-African species from the B. bufo group (i.e. genus Bufo sensu stricto (Frost et al., 2006): B. bufo, Bufo gargarizans and Bufo verrucosissimus), B. viridis group (i.e. genus Pseudepidalea sensu (Frost et al., 2006): B. viridis s.l., Bufo raddei, Bufo brongersmai and possibly the Early Miocene fossil Bufo aff. viridis and Bufo priscus), Bufo mauritanicus (“Bufo” mauritanicus according to Frost et al. (2006)), B. calamita (i.e. genus Epidalea sensu Frost et al. (2006)) and Bufo regularis (i.e. genus Amietophrynus sensu Frost et al. (2006)). Although the separation of the different species within Bufo groups is made according to Frost et al. (2006), we have not retained here their nomenclature (i.e. the new genera). B. viridis sensu lato is here used as its former acceptation before that molecular biology separates it in the western part of its distribution area into Bufo balearicus, Bufo siculus, Bufo boulengeri, B. viridis sensu stricto and Bufo variabilis.

The osteological nomenclature follows Bailon, 1999, Blain, 2005, Blain, 2009 and Sanchíz, 1998. The fossil bones have been illustrated (Fig. 2) using the Environmental Scanning Electron Microscope from the Universitat Rovira i Virgili (Tarragona, Spain) and photographic pictures for larger bones (G. Campeny, IPHES, Tarragona).

B. viridis s.l. as used in this work corresponds today to a widespread group of closely related species with a range that extends from eastern France and Italy to central Asia, including northern Africa and numerous Mediterranean islands (Fig. 4A). Because of the high morphological variability of the green toad, several forms, as species and subspecies, have been described within its extensive range: Bufo balearicus (Apennine Peninsula, Corsica, Sardinia, Balearic Islands), B. siculus (Sicily), Bufo boulengeri (North Africa), B. viridis sensu stricto (central and eastern Europe) and B. variabilis (the Balkans, Anatolia) (Stöck et al., 2001, Stöck et al., 2006 and Stöck et al., 2008). These species today inhabit a wide variety of habitats, from mesic to arid, from subtropical to cold temperate, and from below sea level in Israel to more than 4000 m a.s.l. in the Himalayas (Dessauer et al., 1975). They are common along coasts, due to their ability to survive and breed in brackish waters. Their typical reproductive sites are temporary and shallow water bodies (El Hamouni et al., 2007 and Sicilia et al., 2006).

In the fossil record of western Europe (west of its current continental distribution area, Fig. 4B (Bailon and Hossini, 1990, Rage and Roček, 2003 and Sanchíz, 1998)) the B. viridis group is reported in the Early to Middle Miocene of Spain in Córcoles (MN4), Escobosa de Catalañazor (MN7-8) and Manchones II (MN6) and from France in Vieux-Collonges (MN4) and La Grive-Saint-Alban (MN7 and 8). The earliest extant green toads (B. cf. viridis) are mentioned in the Pliocene of Rincón 1 (MN16, Spain (Alberdi et al., 1982 and Sanchíz, 1998)) and Mas Genegals (MN16/17, France (Bailon, 1991)). As an extension of this work, we consider that two other Spanish Early Pleistocene sites (Barranco León D (Granada) and Almenara-Casablanca 3 (Castellón)) have delivered some remains (mainly ilia) that document the presence of a green toad but whose attribution has until now remained cautiously as Bufo sp. (Agustí et al., 2009, Agustí et al., 2010, Agustí et al., in press, Blain, 2005, Blain, 2009 and Blain et al., in press). The youngest recorded occurrence of the B. viridis group outside but very near to the current westernmost limit of its distribution area was in the Middle Pleistocene of Terra Amata, southeastern France (Bailon, 1991).

A few elements have been attributed to B. aff. viridis in the Late Pleistocene of Cueva Horá (Granada, Spain (Bailon, 1986)), but these probably correspond more exactly to B. calamita (S. Bailon, personal communication). In addition, B. viridis has been mentioned by Holmann (1998) in the Middle Pleistocene of East Farm (Suffolk, England), although Sanchíz (1998) does not mention it.

On the other side of the Mediterranean Sea, in North Africa, B. viridis is known since the Late Pliocene of Ahl al Oughlam (Casablanca, Morocco (Bailon, 2000)).

Various authors have suggested that the climatic conditions linked with the Messinian Salinity Crisis stimulated the propagation of steppe taxa (sensu lato), that these climatic changes were doubtless favorable to the extension from the east of fauna and flora typical of Eurasian steppes, and that dispersals throughout the Mediterranean area may have been greatly favored by the increase in emerged lands due to the partial desiccation of the Mediterranean Sea (Blondel and Aronson, 1999 and Suarez Cardona et al., 1992). In short, it has been suggested that these events at the end of the Miocene provided the environmental basis for the formation of the first Iberian steppes, and that they likewise stimulated the entrance and dispersal of steppe taxa coming from North Africa and the steppe areas of southern Eurasia. According to the molecular data (Stöck et al., 2006), this is probably the case for B. viridis s.l. which suggest a dispersal into western Europe and the western Mediterranean Islands for extant populations prior to the Early Pliocene (around 5 Ma), very probably due to (or favored by) the Messinian Salinity Crisis. Nevertheless a closely related form to B. viridis s.l. (B. aff. viridis) is still present in the Early and Middle Miocene of France (Bailon and Hossini, 1990).

Three scenarios are possible: •

the dispersal of the B. viridis s.l. in western Europe dates from the Latest Miocene (as stated by Stöck et al. (2006)), favored by the desiccation of the Mediterranean Sea, which created new routes for vertebrate fauna. Subsequently its distribution area was progressively reduced and withdrawn to its present-day distribution; in the Iberian Peninsula it survived owing to the particular ecological and/or climatic particularities of the semi-arid landscapes until the end of the late Early Pleistocene in Barranco León D (ca. 1.3 Ma), Cueva Victoria (ca. 1.1–1.2 Ma) and Almenara-Casablanca 3 (ca. 1.1 Ma). According to the study of mitochondrial relationships within the B. viridis s.l. (Stöck et al., 2006), Balearic populations (B. balearicus) are closer to the central European and Asia Minor populations (B. viridis and B. variabilis) than to the North-African B. boulengeri, so that probably the Late Miocene dispersal event came from Asia;

As underlined by Sanchíz (1997), extinction, whether absolute or only in a defined territory, is not probabilistically predictable in relation to ecological or physiological parameters for fossils. In 1989, living B. viridis was mentioned in the province of Murcia (Dicenta et al., 1989), but further investigations have not confirmed its presence. As a consequence, because neither living green toads nor fossils from after the Middle Pleistocene have been recorded from the Iberian Peninsula, it is very probable that B. viridis s.l. became locally extinct in Spain.

During the Middle Pleistocene transition (MPT), the Earth's climatic system is known to have experienced a major change related to intensification of the Northern Hemisphere glaciation, with a significant shift toward more intense glacial conditions, which occurred between 1250 and 700 ka. The emergence of low-frequency, high-amplitude, quasi-periodic (100-kyr) glacial variability during the Middle Pleistocene in the absence of any significant change in orbital forcing indicates a fundamental change internal to the climate system. Its onset was accompanied by decreases in sea surface temperatures in the North Atlantic and tropical-ocean upwelling regions and by an increase in African and Asian aridity and monsoonal intensity (Clark et al., 2006). On lands the MPT corresponds to a major rearrangement of mammalian communities (e.g. Croitor and Brugal, 2010, Cuenca-Bescós and García, 2007, Cuenca-Bescós et al., 2005, Cuenca-Bescós et al., 2010 and Palombo, 2010). Nevertheless, semi-arid steppe environments seem to have occurred in southeastern Spain throughout the Middle and Late Pleistocene, and the climate was not very harsh even during glacial periods (e.g. Carrión et al., 2010 and López-García, 2008). And even if some late Early Pleistocene localities suggest mean annual precipitations to have been higher than today in southeastern Spain (Agustí et al., 2009, Blain et al., 2007, Blain et al., 2008a and Blain et al., in press), a period of aridity is well recorded during summer that favored the existence of semi-arid open landscape (Blain et al., in press). So why did the green toads from the Iberian Peninsula disappear?

Another hypothesis can be made. It seems that today in southeastern Spain the natterjack toad B. calamita has a preference for arid environments and in particular is more likely to be found in areas with very low climatic stability (Romero and Real, 1996). So might the disappearance of B. viridis from the Iberian Peninsula be due to a competition for habitat with B. calamita? Romero and Real (1996) show that in the south of the Iberian Peninsula B. bufo has a different distribution from B. calamita with respect to certain climatic and environmental parameters. Moreover, they suggest that the evolution of these parameters has an influence on the competitiveness of toads. As B. calamita shows the greater adaptability under an unpredictable climate, might it have exerted a pressure on B. viridis leading to its complete disappearance?

B. calamita currently lives in a large part of Europe from the Iberian Peninsula to Byelorussia and Ukraine at the east and Estonia at the north. It is lacking south of the Alps and Carpathian as well as in Mediterranean Islands (like the Balearic Islands for example) (Pleguezuelos et al., 2002). Its earliest fossil occurrence is reported from the Late Miocene of Spain in Concud, Mansuetos (MN12 (Sanchíz, 1977 and Sanchíz, 1998)) and from the Pliocene of Spain and France in Sète (MN15), Islas Medas (MN16) and Montoussé 5 (MN17) (Bailon, 1991 and Sanchíz, 1998). It is noticeable that in Cueva Victoria, Barranco León D and Almenara-Casablanca 3, the presence of B. calamita is very scarce, whereas B. bufo is always represented, though only from a few remains. At the same time, in other Pleistocene localities from northern Spain in the Sierra de Atapuerca, more than 50,000 bufonid fossil remains have so far been studied in the Sima del Elefante (ca. 1.2 Ma (Blain et al., 2010, Cuenca-Bescós et al., 2010 and Rodríguez et al., in press)) and in Gran Dolina (ca. 0.9 to 0.2 Ma (Blain et al., 2008b and Blain et al., 2009)), and no evidence of B. viridis has been found; B. calamita is far less the more abundant and B. bufo varyingly represented but with a much lower abundance.

From an ecological point of view, some characteristics in common between Barranco León D, Cueva Victoria and Almenara-Casablanca 3 can be noted that may have favored the presence of B. viridis over other bufonids. These three localities are currently located in one of the aridest areas of the Iberian Peninsula, with low rainfall (< 400 mm) and high summer temperatures. Moreover, the water available in the vicinity of these localities was very probably brackish (at least temporarily) in the large, shallow lake of Barranco León D (Anadón and Gabàs, 2009 and Gibert et al., 2007), and in Almenara-Casablanca 3 and Cueva Victoria that were very near to the Mediterranean Sea (Fig. 1 and Fig. 4). It is common knowledge that B. viridis is one of the few Mediterranean anurans that tolerates brackish water to reproduce (El Hamouni et al., 2007 and Sicilia et al., 2006), whereas B. calamita usually avoids such environments (Banks and Beebee, 1987 and Gómez-Mestre, 2009). Such competition between B. calamita and B. viridis has never been studied (either in a laboratory or in the field), although it has been tentatively suggested in order to explain the eastern limits of B. calamita (Beebee and Rowe, 2000). Laboratory experiments are not rare for evaluating the competition between genus Bufo toad tadpoles (e.g. Banks and Beebee, 1987 and Katzmann et al., 2003), but again no data are available between B. calamita and B. viridis.

A similar pattern may perhaps have occurred in the replacement of the Mongolian toad (B. raddei) by B. viridis in the East-European Platform. According to Ratnikov (1996), most of the reliable finds of B. viridis in the Russian Platform are restricted to the Upper Pleistocene-Holocene, whereas in the Late Pliocene-Middle Pleistocene B. raddei is found persistently. In the Late Pliocene the Mongolian toad seems to have occupied a vast area including, besides its recent natural habitat, the southern Russian Platform, where it persisted up to the Middle Pleistocene. The green toad, inhabiting eastern Europe (since the Late Miocene [Sanchíz, 1998]), appeared in the Russian Platform only recently in the Late Pleistocene, when the area of the Mongolian toad had been reduced substantially. So far, however, no clear conclusions have been reached on this replacement that may suggest competition between these two toads.

For the first time fossil remains of a green toad (B. viridis s.l.) are identified in the Early Pleistocene of the Iberian Peninsula, in three late Early Pleistocene localities: Barranco León D (ca. 1.3 Ma), Almenara-Casablanca 3 (ca. 1.1 Ma) and Cueva Victoria (ca. 1.1–1.2 Ma). They correspond to the last occurrence of the B. viridis group in the westernmost part of Europe.

Since no osteological differences have been established between the recently described extant species of B. viridis s.l. (e.g. B. balearicus, B. siculus, B. boulengeri, B. viridis and B. variabilis) no precise palaeobiogeographical relationships can be drawn for the Spanish fossils. However, the possible (although improbable) entrance of these toads via the Strait of Gibraltar is of particular relevance in the debate on the post-Messinian North Africa-Europe interchange. Moreover, the occurrence of a third species of bufonid toad during the Pleistocene in the south of the Iberian Peninsula raises some interesting ecological questions in relation to the local disappearance of the green toad, which can be hypothetically linked with the intensification of the Pleistocene glacial/interglacial climate dynamics (emergence of low-frequency, high-amplitude, quasi-periodic 100-kyr glacial variability) and/or with probable competition with another toad, B. calamita, which today in southeastern Spain shows a preference for arid environments and in particular is more likely to be found in areas with very low climatic stability.

As perspectives opened up by this work: an osteological revision of all the newly-named species of the B. viridis group is required in order to establish the relationship of the Spanish fossils with extant species; also, a revision of the bufonid fossil record in Spain since the Miocene and in particular in the Late Pliocene – Early Pleistocene is needed in order to clarify whether there were no, one or two dispersal events for green toads in western Europe.