The analysis performed using traditional micrographic techniques (De Palacios, 1997 and Schweingruber, 1990) of the 6 woody fossil remains revealed the following diagnostic features:

Description. Transverse section: homoxylous wood with perfectly delimited growth rings and polygonal tracheids with an approximate diameter of 28.6 μm. Abrupt transition from early (average of tracheids 34 μm) to the late wood (average of tracheids 20 μm). Physiological longitudinal resiniferous channels (average 102.2 μm) preferentially located (93.2%) in the summer wood and transition zones (Fig. 2, A, D). Radial section: ray-tracheids with rarely isolated dentations (only 20%). Most of these are sharp (80%), although they can also be found in varying degrees of bluntness (Fig. 2, B). The length of the dentations is variable (average 6.27 μm, up to 12 μm), reaching the centre of the lumen on several occasions, and even interconnecting. Cross-fields from parenchyma cells to tracheids with one large window-like pit (average 22 μm), sometimes 2 (Fig. 2, C). Spiral thickenings absent. Tangential section: heterogeneous woody rays uniseriate or pluriseriate, generally short, up to 9 cells, (Fig. 2, E), with the presence of some pluriseriate ones in which transversal resiniferous channels can be observed. Areolate pits on the radial walls of the longitudinal tracheids arranged in uniseriate rows, in no case biseriate.

Identification. This homoxylous wood with resiniferous channels, ray-tracheids walls with well-developed teeth and one or two window-like pits is characteristic of the genus Pinus, in fact of the P. sylvestris group including P. sylvestris L., P. nigra Arnold and Pinus uncinata Ramond ex DC. (Greguss, 1955 and Schweingruber, 1990). Some features observed were quite different from those of P. uncinata. On one hand, the diameter of the channels was much greater than that established for the latter taxon (average 80–90 μm), and their preferential (92.3%) location in summer and transition wood does not coincide with the preferential location of these elements in spring wood in P. uncinata (García and Guindeo, 1988 and Peraza, 1964). On the other hand, we did not observe remains of epithelial cells after the preparations had been made. The characteristics of the dentations of the ray-tracheids are among the most commonly mentioned diagnostic criteria in the literature (De Palacios, 1997, García and Guindeo, 1988, Greguss, 1955, Jacquiot, 1955 and Peraza, 1964). The height of the dentations was frequently over half the diameter of the lumen, both characteristics of Pinus sylvestris leaving columnar formations (De Palacios, 1997 and Greguss, 1955). Approximately 80% of the projections were pointed, compared with 20% which were blunt, once again features established for P. sylvestris (García and Guindeo, 1988, Greguss, 1955, Jacquiot, 1955 and Peraza, 1964). Finally, 80% of the dentations were not isolated, further suggesting P. sylvestris (García and Guindeo, 1988 and Peraza, 1964).

Studies with reflected light microscopy

Description. Homoxylous wood with longitudinal tracheids and the presence of pits on the radial walls arranged in uniseriate rows. Uniseriate rays (some pluriseriate ones), generally short (number of cells lower than 12s). Cross-fields from parenchyma cells to tracheids with one or two large window-like pits. Identification. From the set of characters observed, we only deduced that all the wood fragments analysed correspond to the genus Pinus, and within this, to some of the species: P. sylvestris, P. nigra or P. uncinata (Greguss, 1955 and Schweingruber, 1990).

All the wood samples can therefore be identified with certainty as pertaining to the Pinus sylvestris group. Taking into account the set of characters described and defended by some authors (De Palacios, 1997, García and Guindeo, 1988, Greguss, 1955, Jacquiot, 1955 and Peraza, 1964), the 6 samples analysed exhibit characters that coincide with the wood described for the taxon P. sylvestris. However, the difficulty involved in differentiating among the wood fragments of the group, particularly on considering diagnosis with small-sized fragments, with no possibility of avoiding early wood (Figueiral and Carcaillet, 2005 and Mutz et al., 2004), suggests to assign tentatively our samples to the taxon P. cf. sylvestris.

Making taphonomic considerations of the wood samples from the Cerralbo collection clearly presents limitations resulting from the scant data regarding the conditions thereof at the site of extraction. It is not a simple task to differentiate the different organs of the wood into twigs, branches or trunks (Chapman, 1994). The great similarity in anatomical microstructure and the natural variability in the wood resulting from seasonal, climatic and developmental variations prevents easy adscription to one or another anatomical component. All the wood fragments from the Cerralbo collection possess very narrow sections, with rings very close together, which would likely indicate quite thin fragments of twigs or branches. However, we cannot rule out the possibility that some of the fragments could correspond to disarticulated remains of trunks that were broken down into small components due to the disintegration process during the biostratinomic phase. Neither can we rule out the possibility of occasional fragmentation during the extraction processes. In all cases the remains possess no bark, which might have been lost during the dispersal and biodegradation processes.

Although we cannot provide an accurate consideration regarding the possible displacements of the Torralba wood fragments, studies conducted on macroremains indicate displacements of a maximum of only 10 km in continental environments, which would appear to indicate conditions of at least a relative regional autochthony (MacGinitie, 1969 and Rich, 1989). In the wood fragments from the Cerralbo collection, there are no signs of wear or abrasion. In most cases, they present surfaces with the typical irregularities found in wood fragments, and caused by natural fragmentation (natural death of the parent tree, breakage by the weight of snow, animals, etc.) and it could be stated that their dispersal was very likely caused by simple flotation (Gastaldo, 1986 and Gastaldo, 1989). We therefore believe that it cannot be said that these wood fragments were manipulated by man, as suggested both by Cerralbo (1911) and by Howell et al. (1962).

The state of the wood fragments has remained essentially unaltered since their original composition; we found no evidence of carbonification and most of them preserve their original colouring and weight. This preservation most likely results from early burial of the fragments which, under anoxic condition, prevented biodegradation processes. In some of the wood fragments, however, there is evidence of carbonization on the outside. This is the case of samples: without number (ref. 75/24/61 and showcased at the Archaeological Museum), n^o 2751 (ref. 75/24/58), n^o 2724 (ref. 75/24/43) and n^o 2722 (both fragments). To these must be added another 10 contained in bags with no reference at all. Considering the sedimentary environment in which the wood fragments were found, we believe that the carbonization must have occurred during the biostratinomic phase. Analysis of these wood fragments did not provide us with sufficient elements to discern whether these carbonization processes are anthropic or natural. With the data available, we cannot establish whether they are all from the same parent tree or from different individuals. In the fossil diagenetic phase, lithostatic pressure flattened the pieces and caused cracks and fractures in most of them. Neither did we find signs of sedimentary infilling that might indicate processes of taphonomic reworking.

Diagnosis of the wood samples from the excavation by the Marquis of Cerralbo reveals the presence in the centre of the Peninsula of the taxon Pinus cf. sylvestris for the Middle Pleistocene. This finding confirms the results of previous pollen analyses performed at the site, which indicate that the P. sylvestris typology makes its appearance therein (Menéndez Amor and Florschütz, 1959). According to these authors, these analyses reveal a set of hygrophilous or mesohygrophilous taxa such as Salix, Corylus, Alnus, Betula together with the so-called Quercetum mixtum. The appearance of these taxa might be related to valley-bottom environments, or shaded ones and those presenting a high phreatic level. Although on occasions the pines could be associated with more of less hygrophilous environments, they generally occupy large areas on the surrounding moorlands, constituting more or less open formations comprising quite high percentages of Gramineae and Chenopodiaceae (Menéndez Amor and Florschütz, 1959). It was subsequently pointed out that these pine pollen percentages were variable, although an invariable number of species represented was typically observed (Freeman and Butzer, 1966 and Howell et al., 1962). The most recent and complete palynological study in the area was performed by Ruiz-Zapata et al. (2003) at the Loma del Saúco; it presents two pollen levels, C and B. At pollen level “C” or below, situated at a depth of between 100 and 40 cm, a dominance of tree pollen can be observed, with Pinus and Juniperus mainly represented, together with the more or less continuous presence of Castanea, Juglans, Corylus, Alnus and, to a lesser extent, of Betula, Fagus and deciduous Quercus. At the next level, “B”, at a depth of between 40 and 20 cm, there is a noteworthy increase in Juniperus and a simultaneous decrease in Pinus pollen, along with the disappearance of the mesohygrophillous species, except for Betula, a fact that has been associated with an evident drop in temperature and moisture in the area (Ruiz-Zapata et al., 2003).

From an ecological point of view, the appearance of the species P. sylvestris fits perfectly within the floristic assemblage described, both due to its hydric requirements (it grows in valley-bottoms or on slopes) and to its tolerance to cold periods, which must have occurred, judging by the palaeovegetation described. At the present time, the nearest villages are those situated in the Ministra Mountains, at an altitude of 1309 m asl (Bermejo and Martínez-García, 1997), and at an elevation of 200 m higher than the site, which provides clear evidence of its presence at clearly lower elevations than the present time for the Middle Pleistocene represented in Torralba.

The nearby site of Ambrona (Pérez-González et al., 1997a and Pérez-González et al., 1997b) shows certain similarities with Torralba. According to Howell et al. (1995), two associations appear in the so-called Lower Complex. On one hand, the Pinus-sedge association, characterised by a high frequency of tree pollen (mainly pine 75–79%) and reeds (up to 10–25%). On the other, the Gramineae–Chenopodiaceae association, characterised by AP values (40–60%), less than 5% of Juncaceae and a high frequency of gramineae (20–35%) and Chenopodiaceae (3% or over). Grasses and chenopods, with a wide range of herbaceous species, were dominant in the valleys and highlands (NAP usually 40–90%). Some tree taxa appear, such as willows and oaks, but very poorly with regard to pollen percentages. Furthermore, in the so-called Upper Complex, Howell et al. (1995) describe a variant of the Pinus–Juncaceae association, with an AP of 96–98% and some rare deciduous species like oak (1.5%) and elm (0.7%). Once again, a moderately cold and moist glacial phase appears, supported by geomorphological evidence. The most recent study in Ambrona by Ruiz-Zapata et al. (2003) highlights in its four pollen zones the presence of vegetation dominated mainly by Pinus, Juniperus and Poaceae, with the presence of mesohygrophilous taxa and riparian elements such as Castanea, Corylus, Juglans and Ulmus with the occasional presence of Betula. Mediterranean taxa such as Olea or evergreen Quercus also appear, with abundant Poaceae and continuous high percentages of Ericaceae and Rosaceae. Throughout the whole diagram, one can appreciate different phases of decreases or changes in temperature and moisture, perhaps associated with local variations in hydric availability, which in turn generate local variations in the percentages of mesohygrophilous elements. In the final zone, pollen zone IV, one can observe the development of a monospecific forest of Pinus, the subsequent degradation of which gives rise to a phase of almost exclusive dominance of Juniperus, with wide-ranging steppic elements and a decrease in aquatic species, values indicative of a drop in mean humidity and temperature in the basin (Ruiz-Zapata et al., 2003).

The appearance of the Pinus sylvestris group (P. cf. sylvestris) at the Torralba site constitutes the first evidence based upon wood fossils of this taxon for the Middle Pleistocene on the Iberian Peninsula. The only macroremains previously referring to this taxon are some charcoals attributable to Pinus sylvestris-type, with an age of over 300 ka, from the Cuesta de la Bajada site (González-Sampériz et al., 2010). Together with Torralba, only the Mealhada and Villaverde sites (Andrade, 1944 and Menéndez Amor and Florschütz, 1959) presented the P. sylvestris pollen typology have been previously cited for the Iberian Peninsula for this part of the Pleistocene.

Several sites, however, reflect the presence of pines and more or less open pine forests, during the Middle Pleistocene in which there are varying degrees of steppic conditions. In the central-north of the Peninsula, at the Atapuerca site (Galería, pollen zone IV) (Fig. 3, B), dated between 256–200 ka (Berger et al., 2008), a phase can be detected in which Pinus, Asteraceae (Liguliflorae) and Poaceae are well represented, with high percentages of non-tree pollen (García Antón and Sainz Ollero, 1991). Deciduous and evergreen oaks practically disappear, as do the Mediterranean taxa present in other zones of the diagram, like Pistacia and Olea. Nevertheless, the site as a whole does not appear to reflect conditions that are too xeric and/or steppic. For the centre of the Peninsula, Menéndez Amor and Florschütz (1959), describe the flora of the Villaverde site (Fig. 3, C), the age of which, according to the authors, is similar to that of the Torralba and Ambrona sites, for which the existence of pines has been mentioned, in a steppic environment with the presence of taxa such as Quercus or Juglans. In Villacastín (Fig. 3, D) dated as MIS 6, open-mosaic vegetation appears with the presence of tree taxa such as Pinus or Juniperus with more or less steppic open zones, herbaceous vegetation and patches of Quercus (Carrión et al., 2007). The last notable site in the centre of the Peninsula is the so-called Pinedo Formation (river Tagus) (Fig. 3, E), which also presents a mosaic-like vegetation with Pinus and Quercus with the presence of both Mediterranean taxa (e.g., Olea, Cistaceae) and temperate ones (e.g., Betula, Castanea, Alnus) (Martín-Arroyo et al., 1995 and Ruiz-Zapata et al., 2004).

In the Northeast of the Peninsula, one can mention the Mindel–Riss Pla de les Presses site (Fig. 3, F), where a somewhat similar set of species can be found, with Pinus, Betula, Artemisia and Quercus, all these taxa also being present in Torralba (Pérez-Obiol et al., 1986). The Val d’en Bas (Fig. 3, F) site, dated at over 250 ka, also presents a floristic assemblage in which Pinus and Abies appear together with mesophytes like Quercus evergreen, Carpinus, Tilia or Corylus, all of these associated with a phase of the interglacial (Cros et al., 1986). Finally, the Northeast of the Peninsula exhibits the Pla de l’Estany site, dated between 190–170 ka (Fig. 3, F); it also registers a cold steppic phase with the spread of Pinus, Poaceae and Asteraceae (Burjachs, 1990). Also? in the North-East of the Peninsula, the Cuesta de la Bajada site (Fig. 3, G) provides similar data, showing steppic conditions with the presence of macroremains of Juniperus thurifera and P. sylvestris-type for an age > 300 ka (> MIS 9) and conditions that are also steppic in an open landscape for MIS 6 dominated by herbaceous species and with the isolated presence of Pinus (González-Sampériz et al., 2010 and Santonja et al., 2000).

In the South-East of the Peninsula, the Padul sequence (Fig. 3, H) includes in its base the Mindel–Riss interglacial and the Riss interglacial, also showing steppic phases in which cold open landscapes appeared to alternate with temperate phases (Florschutz et al., 1971). Similar results were obtained from the marine sequences, for which there is a good general correlation with the continental ones. In the Southwest of the Peninsula, the sites of uncertain age (Mindel–Riss/Riss) of Mealhada (Fig. 3, I), San Pedro de Muel, Ribeira do Penegral and Alpiarça (Fig. 3, J), Porto Covo Sines (Morgavel), Ribeira da Provença, Samourqueira and Vila Nova de Milfontes (Fig. 3, K) (Andrade, 1944, Teixeira, 1943 and Zbyszewski, 1958) present a microthermal and/or hygrophilous set of species in which Pinus, Betula, Ericaceae and Quercus frequently appear. At all these sites, pine species belonging to the Pinus sylvestris group might have made up a part of their communities, despite being close to the coast and far from their current distribution ranges. However, the possibility cannot be ruled out that Pinus pinaster also formed a part thereof. Thus, the marine records showing the presence of Pinus (e.g., the MD95-2042, MD01-2447, MD01-2443 cores in Desprat et al., 2006, Roucoux et al., 2006 and Sánchez-Goñi et al., 1999) (Fig. 3, L) coincide in general terms with the data presented by the continental sequences, highlighting the fact that in the colder phases of the Middle Pleistocene, steppic or pseudo-steppic formations with a presence of pine trees were widespread.

Despite the difficulty involved in dating the European sites of the Middle Pleistocene, there is an apparent correlation between the data from Torralba and those obtained at other sites of a similar age in SW Europe. An alternation of cold and warm phases during the Middle Pleistocene has been indicated in numerous sequences for the whole continent. For southern Europe, we found certain parallelisms in some sites where it is possible to find the set of species present in Torralba. Thus, cold, arid periods characterised by open vegetation have been highlighted at the Vallo di Diano site in southeast Italy, for periods of the Middle Pleistocene falling within isotopic stages 16 and 14 (Russo Ermolli and Cheddadi, 1997). In Italy, too, in the Valle di Castiglioni sequence, one can appreciate similar steppic conditions and phases of approximately 250.000 BP (Follieri et al., 1988). In southeastern France, the Terra Amata site, associated with the Torralba and Ambrona sites due to its lithic industry, also indicates the presence of Pinus sylvestris-type during this phase of the Pleistocene (Mourer-Chauviré and Renault-Miskovsky, 1980). Accompanying this species are Alnus, Salix, Betula, Fagus, Picea and Abies. These authors state that the non-Mediterranean taxa likely reduced their elevation as a result of changes in climatic conditions such as cold and moisture, often coming close to the coast. This type of vegetation for the Middle Pleistocene can also be found in the Greek sequence of Tenaghi Philippon, where one can observe that in zone 26 (isotopic stage 12) steppic conditions appear with a phase of Artemisia-Chenopodiaceae showing different values of Pinus and different deciduous species (Van der Wiel and Wijmstra, 1987). Similar situations occur at the site in zones 32, 30 and 28, where different steppic situations can be observed.

The appearance of P. cf. sylvestris at the Torralba site constitutes unequivocal evidence of the most extensive past distribution of montane pines during cold periods of the Pleistocene. Indeed, numerous fossil macroremains testify to these decreases in altitude of montane pines during the Upper Pleistocene or even the Holocene (Badal et al., 2013 and Postigo-Mijarra et al., 2010). Interestingly, these notable decreases in elevation of P. sylvestris or P. nigra involved areas relatively close to the coast, thus testifying to a clear altitudinal descent of these species during the coldest periods of the last glacial–interglacial cycle. The presence of these species at lower elevations than the present is now evidenced by the existence of natural populations of P. sylvestris to be found today in certain parts of the Duero basin, such as the Cega River (Segovia), and at higher elevations, pine forests in Lillo (León), Velilla del río Carrión (Palencia), the Nevada, Cazorla, Baza and Gêres ranges (Costa Tenorio et al., 1997, Franco Múgica et al., 2000 and Franco Múgica et al., 2001).