This study includes an analysis of small mammal remains from 9 samples from sublevels TE9a, TE9b and TE9c. The process of microfossil recovery begins with the washing and screening of all the sediment obtained from the archaeo-paleontological excavation of the Sima del Elefante site. The washing is done with three superimposed sieves of decreasing wire mesh size (5 mm, 2 mm and 0.5 mm). Once the concentrate produced by the washing has been dried, the sorting takes place. Later, each of the remains is analyzed through a binocular magnifying glass (OPTECH SL Dual 6.5 ×–60 ×). For a detailed study of the remains and high-resolution photographs, a variable-pressure environmental scanning electron microscope or ESEM (FEI QUANTA 600) is used under low-vacuum conditions, applying a high voltage (15 to 20 kV) and using secondary electron detectors (LFD detector) at the University Rovira i Virgili of Tarragona (Spain).

Given the large quantity of material we have available, we use selected samples from the same squares throughout the sequence TE9 always proportional to the excavated volume of sediment and the density of bone within it. From each of the samples all the remains are recorded, and a taphonomic analysis of the rodent and insectivore elements is performed. Lagomorphs are not included and will be the subject of a future taphonomic study. Order, family and species are identified using M₁ as a diagnostic element, but in the absence of this element a diagnostic cranial or postcranial element is used. The determinations are based on the works of Cuenca-Bescós et al., 1995 and Cuenca-Bescós et al., 2013, Laplana and Cuenca-Bescós (2000), Rofes and Cuenca-Bescós, 2006, Rofes and Cuenca-Bescós, 2009a, Rofes and Cuenca-Bescós, 2009b and Rofes and Cuenca-Bescós, 2013. Moreover, the fossils were grouped using the minimum number of individuals (MNI) method, from each sublevel, by counting the best diagnostic element, taking into account, whenever possible, laterality.

For the taphonomic study of the remains, the systematic-descriptive method described by Andrews (1990) is used. Superficial alterations present in the small mammal remains are observed and described (the representation of skeletal elements, the breakage and the degree of digestion), allowing the intervention of the predators to be identified and recognized.

To understand and describe the representation of the elements, we use indicators of the relative abundance of the skeletal elements (Andrews, 1990) and of the proportions of the skeletal elements (postcrania/crania; femur + humerus/maxilla + mandible; radius + tibia/humerus + femur; following Andrews, 1990 and Fernández-Jalvo and Andrews, 1992). Subsequently, the fractures and the digestion of the elements are analyzed differentially and specifically for the postcranial elements, maxilla, mandible, molars and incisors (Andrews, 1990 and Fernández-Jalvo and Andrews, 1992). Finally, the study is completed with an analysis of the postdepositional modifications present on the surface of the fossils. Whenever possible, these modifications are recorded taking into account the degree of alteration, although in the majority of cases only their presence or absence could be noted.

The total number of fossils studied from TE9 of Sima del Elefante is 4784, with a total MNI of 491. TE9c is the sublevel with the fewest remains and the lowest number of species (Table 2 and Table 3). All the levels feature individuals belonging to the class Mammalia, especially the orders Rodentia and Eulipotyphla (Table 2). Among the rodents Allophaiomys lavocati stands out, as does Talpa cf. T. europaea within the order Eulipotyphla (Table 2).

Among the sublevels of TE9, there are similarities in the presence of species and the alterations. All the anatomical elements are present in each sublevel (Table 3). The relative abundance of the skeletal elements indicates that the average representation is 32.9%, with sublevel TE9b remarkable for its low average representation (25.3% of the remains expected for the MNI calculated). The ratios between postcranial and cranial elements show a predominance of cranial remains (Table 4). Notable is the preferential destruction of the distal elements of the legs (Table 4). The breakage of the elements in TE9 and the loss of teeth in maxillae and mandibles are high (Table 5). The TE9b assemblage, despite its low representation, shows slightly lower fracture rates than TE9c and TE9a, where the breakage is mostly heavy and extreme. Note, for example, that the molars in the samples from TE9b tend to be retained in their alveoli and that there are more complete postcranial elements and mandibles than in the other sublevels (Table 5).

In all the samples studied, there are elements of rodents and insectivores with signs of digestion (Fig. 2). The digestion in levels TE9c, TE9b and TE9a is similar and affects about 30% of the remains (Table 6). In all sublevels, the most frequent degree of alteration is light (Table 6), although it is noteworthy that heavy degrees are reached in incisors (and even extreme degrees in TE9b) and moderate degrees in molars and postcranial elements.

Regarding postdepositional alterations, there is abundant precipitation of manganese oxides in TE9, descending slightly in TE9c (Table 7). Alterations caused by water abrasion as well as by changes in humidity and dryness have been recorded, including concretions (mostly in TE9c), cracking and fractures produced by fallen rocks and sediment pressure. The differences between the three sublevels are present in alterations affecting less than 2% of the remains (dissolution, chemical corrosion, weathering and grooves caused by roots). It is worth highlighting that only in TE9b have we recorded cracking and exfoliations related to subaerial exposure and only in TE9a are there alterations caused by roots (Table 7).

The results of the taphonomic study of the small mammal remains from unit TE9 of Sima del Elefante shed valuable light on the landscape of the Sierra de Atapuerca and on the depositional microenvironment from 1.2–1.3 Ma.

In the light of the alterations caused by digestion, it has been estimated that the origin of the small mammal accumulations is related to the biological activity of predators. While these alterations appear to be similar among the three sublevels, detailed analysis reveals certain differences. It is thus observed that the small mammal fossils from TE9b show less fracturing but at the same time higher degrees of digestion than those from TE9c and TE9a, which are more similar to one another.

In TE9c and TE9a, the percentage of molars and incisors affected by digestion is greater than expected for a category 1 predator of modification (0–6% molars, 8–30% incisors), but slightly lower than expected for a predator category 2 (11–22% molars, 48–70% incisors) (Andrews, 1990). However the percentage of postcranial digested (25–50%) and degrees of alteration achieved in molars and incisors (moderate and heavy digestion) could correspond to that exerted by a category 2 predator of modification (Andrews, 1990). Among the predators in this category are the spotted eagle owl (Bubo africanus), the eagle owl (Bubo bubo) and the tawny owl (Strix aluco). The alterations produced by these predators are similar, and it is difficult to distinguish small mammal assemblages made by either bird of prey from those of the other. Further, the percentage of postdepositional breakage (fallen rocks, sediment compaction) is very high in all the sublevels and has obscured the original predator-induced breakage patterns and representation percentages of bones. This could be responsible for the fact that the representation profiles do not match the proposed predators (Fig. 3).

Nevertheless, in current assemblages, small differences can only be seen in the intensity of alteration (especially in fractures and digestion, which are slightly higher in the tawny owl) or the overall size of the prey consumed (which is usually greater in the eagle owl) (Andrews, 1990). First, the degrees of digestion observed in the remains from TE9c and TE9a are higher than expected in remains consumed by an spotted eagle owl (reaching moderate degrees in molars) and eagle owl (reaching heavy degrees in incisors) (Table 6), suggesting that these accumulations are formed by the tawny owl. Secondly, in concentrations with the eagle owl it is common to see skeletal elements belonging to Leporidae, weasels and small herbivores (Andrews, 1990, König et al., 1999 and Mikkola, 1983). In sublevels TE9c and TE9a there are abundant leporid remains. These leporids show a good representation of all the skeletal elements and generally have low fracture levels (Huguet, 2007). Furthermore, some of these elements, mainly limbs, have been found in anatomical connection (Huguet, 2007). Taphonomic studies of these taxa have not identified surface modifications caused by consumption by predators, such as bites, pecks or digestion marks (Huguet, 2007). Therefore, it is possible to rule out the intervention of an eagle owl on these remains, since one would expect a greater presence of diaphyseal cylinders, peck marks and epiphyses with light or moderate signs of digestion (Andrews, 1990). All this seems to indicate that within predator category 2, the likely predator is the tawny owl. Besides this raptor, the intervention of Beremendia fissidens (Early Pleistocene fossil shrew) as a sporadic predator on a mole (Talpa cf. europaea) humerus was also shown in TE9a (Bennàsar et al., 2015).

By contrast, the degrees of alteration in sublevel TE9b do not correspond to those produced by any known predator. Thirty seven percent of the incisors show signs of digestion, which would indicate a nocturnal raptor and certain viverrids as the authors of the assemblage. However, this digestion attains extreme levels of alteration, often associated with predation by diurnal raptors. Furthermore, and despite having a lower average of relative abundance of the skeletal elements, the assemblage from TE9b shows slightly lower percentages of fracture and tooth loss than those from TE9c and TE9a, where a nocturnal raptor has been identified as the author.

For this assemblage two possible hypotheses may thus be put forward: 1) It could be a predator of an intermediate category with an unusually high degree of digestion, or 2) it could be a mixture of predators. In the first case, it should be noted that several factors are involved during digestion, leading to variability in the alterations produced by different predators, both at the species and the individual level. One of these variables is the pH and enzyme level of the predator (Smith and Richmond, 1972). The pH is more acidic before eating and when the predator has gone for a long period without eating any prey (Duke et al., 1975). This factor, along with the time spent in the stomach, can produce more or less aggressive effects on the remains of small mammal concentrations (Chitty, 1938). Furthermore, the pH is increased in young predators (Andrews, 1990. Raczynski and Reprecht, 1974). Other factors that determine the degree of alteration are the position of the remains in the stomach and the existence of hair and feathers enveloping the skeletal remains for regurgitation, etc. (Andrews, 1990).

The possibility of a mixture of predators should also be taken into consideration. The mixture of the assemblages contributed by various predators can take place either during the production of the pellets/depositions (that is, there are depositions from several predators in one place before burial) or during fossilization. This sublevel has not revealed any sedimentological, chronological or faunal changes that might suggest the presence of linear changes or overlaps, or any alteration that might indicate that the mixture was produced by a reelaboration of the elements, thus ruling out the possibility that the mixture is postdepositional in origin.

A previous experiment where current assemblages of known predators were mixed (Bennàsar, 2010) has suggested that mixtures can be detected when there is a majority predator in them (one that provides about 70% of the remains in the assemblage), and in some cases, it is possible to identify the species in question. According to this study, one of the main diagnostic parameters for identifying these associations is the presence of discrepancies between the degree of digestion of the incisors, molars and postcranial elements. Moreover, this disparity can also be observed in the levels of general alteration of representation and fracture. However, it should be borne in mind that in archaeo-paleontological assemblages these parameters can be modified by fossil-diagenetic agents.

In the fossils from TE9b, the degree of digestion presented by the molars, postcranial elements and, above all, the incisors shows discrepancies, which would support the hypothesis of a mixture of predators. Within this possible mixture, on the basis of the main alterations observed, a category 2 predator would be identified as the principal agent, probably the same predator as in sublevels TE9c and TE9a (tawny owl). In addition to this predator, there may be the sporadic presence of a predator with higher degrees of alteration or a category 3 predator, reflected in the low representation and in the extreme digestion present in the incisors, although it would not be possible to specify the species to which it belongs.

Thus, the predator responsible for the accumulation of remains from TE9c and TE9a is the tawny owl (Strix aluco), which is possibly also the main predator of TE9b (Table 8). This small bird of prey with opportunistic, nocturnal habits includes rodents, voles, shrews, dormice, small birds and insects in its diet. They are strongly territorial and sedentary birds. They nest on rocky walls or in tree hollows, using pellets for quilting eggs. As regards their preferential habitat, they opt for semi-open forests, generally of leafy, deciduous trees and mixed with clearings. They are also common in riverside forests, open landscapes near forests, and in rocky areas with scattered trees and bushes (König et al., 1999, Lipej and Gjerkes, 1996, López-Gordo, 1974, Mikkola, 1983 and Southern and Lowe, 1968). We can thus infer a landscape with semi-open riverside forests with nearby open areas in the Sierra de Atapuerca during the formation of TE9.

These data are consistent with those provided by the taxonomy and analysis of the diversity of the small mammals, herpetofauna, avifauna and macromammals. The paleoecological studies undertaken to date on the basis of rodent and insectivore taxa suggest the existence of a fluvial or lacustrine landscape and a succession of warm and cold climatic events (Cuenca-Bescós and Rofes, 2004, Cuenca-Bescós et al., 2004, Cuenca-Bescós and García, 2007, Cuenca-Bescós et al., 2013, Rofes and Cuenca-Bescós, 2006 and Rofes and Cuenca-Bescós, 2013). In addition, archaeobotanic and faunal results indicate the nearby presence of meadows, possibly humid, semi-open or open, and of semi-open forests (Rodríguez et al., 2011, Rosas et al., 2001 and Rosas et al., 2006). The area near the Sierra de Atapuerca from 1.2–1.3 Ma thus presented a biome consisting of semi-open riverside forests, or large areas of more or less permanent water, with open meadows nearby (Blain et al., 2010). The climate would probably have been mainly temperate (Rodríguez et al., 2011).

As noted, the predators identified in TE9c and TE9a are opportunistic, and the species found in their depositions are a reflection of the community alive at that time, given that the predator is limited only by the size of the prey and its availability (Table 8). Thus, paleoecological interpretations based on the relative abundances of the small mammal taxa are a good reflection of the hunting habitat of the predators that consumed them. However, the mixture of predators alluded to in TE9b does not allow for this sort of inference.

The microenvironmental characteristics of unit TE9 have been identified through the postdepositional alterations observed in the fossils. These indicate similar conditions for the entire sequence studied. Thus, all the sublevels have been found to show manganese oxide depositions, abrasion, cracking and concretions, although with varying degrees of modification. Other alterations such as dissolution, grooves produced by roots, desquamation weathering and chemical corrosion exhibit a sporadic and uneven presence in the sublevels.

Most of the alterations found are closely related to the microenvironment of the cave, characterized by the low presence of light and by high humidity. These cave conditions are also evidenced by the non-pollen palynomorphs, mainly algae and fungi, which are present above all in TE9 (Rodríguez et al., 2011).

As regards the production of the remains, the alterations identified indicate that they are resedimented. Although the tawny owl can nest or rest on rocks and rocky soils, in shelters or small blastings (Andrews, 1990 and Mikkola, 1983), the alterations shown by some remains (e.g. the abrasion and weathering) indicate that they were not produced in the cave, but were subject to a process of transport from the initial place of deposition located at one of the entrances to the cave.

Such transport may be produced by various mechanical processes, although the abrasion present in all the sublevels of TE9 indicates a predominance of mechanisms associated with water. In TE9c and TE9a, rounding by abrasion affects a small percentage of the remains and generally appears only in isolated cases among the remains, while in TE9b this alteration is more abundant and is sometimes linked to polished surfaces. Such modifications are not abundant or common in small mammals. Although it has been established experimentally that the lower degree of abrasion would occur with silt and clay or fine sand in suspension on fresh bones for a short period of time (Fernández-Jalvo and Andrews, 2003), the suspended-load transport of the remains might not make any alterations. In addition, for this type of transport one should also take into account the partial or total organic covering of the remains at the time of transport and the degree of cohesion of the pellets, because these factors would influence the presence and degree of modification. Thus, the presence of all the skeletal remains in the samples from Sima del Elefante, including the smaller ones, could be associated with the entry of more or less cohesive pellets, and only the remains that were exposed to the erosive particles would present alterations caused by abrasion. The scarcity of postdepositional changes, such as weathering and root grooves, corroborates this hypothesis.

The existence of humid environments at different times during fossilization is evidenced by the intense bacteria and fungi activity observed (Courty et al., 1989, Dorn and Oberlander, 1981, Marín-Arroyo et al., 2008 and Marín-Arroyo et al., 2014). Manganese oxide depositions appear on the small mammals in all the sublevels, and are also observed on the surface of the fossil macrovertebrates from Sima del Elefante (Huguet, 2007), suggesting further that the fossilization process of both types of remains occurred in the same microenvironmental conditions. Some fossils exhibit cracking, dissolutions, and concretions, indicating a humid environment with dripping followed by periods of desiccation and temperature changes. All this suggests that the highly humid endokarstic conditions, possibly with flooding, were not favorable for the development of human occupations inside during the formation of TE9. However, the fact that in TE9c the manganese decreases slightly and the concretions increase indicates a slight reduction in humidity and longer periods of desiccation, during which hominids could have taken advantage of to strengthen their presence in the vicinity of the cave. This increased anthropic presence would explain the decrease in the number of small mammal remains and the MNI in this sublevel (the greater the human presence, the less the presence of birds of prey). This is further supported by the greater number of lithic artifacts recovered in TE9c (de Lombera-Hermida et al., 2015).

Our taphonomical study of small mammals indicates that the associations from TE9c and TE9a are the result of predation by the tawny owl (Strix aluco). By contrast, the small mammal remains from TE9b may have their origin in a mixture of several predators (including the possible participation of the tawny owl and a category 3 predator). These raptors have opportunistic habits, so their concentrations can be considered a reflection of the representation of the species present in the ecosystem. This facilitates paleoecological interpretations based on the relative abundance of the small mammal taxa of TE9c and TE9a.

Our identification of the predator suggests that during the formation of TE9 (1.2–1.3 Ma) the area near the Sierra de Atapuerca had a biome consisting of semi-open riverside forest or large areas of more or less permanent water, with open meadows nearby.

The origin of the remains is mainly exokarstic, and transport into the cave was carried out by various mechanisms of a hydrological nature when the pellets still contained organic matter that covered the remains. The endokarstic environment during the formation of unit TE9 is characterized by high humidity and ponding, preventing the occupation of the cave by the hominids that lived in the vicinity of the Sierra de Atapuerca. Nevertheless, it has been observed that these endokarstic conditions became slightly less harsh in TE9c, which favored their increased presence in the area near the cave.