Introduction

There are different approaches to the mobility models of hunter-gatherer groups, of which the one defined by Binford (1980) is the most widely used. Broadly speaking, two typologies are defined that respond to the type of organisation of the territorial exploitation by the groups, the forager and the collector models. The forager model takes place in areas of demographic dispersion. These are groups that are characterized by a high mobility between several nearby settlements, and they exploit resources in very small areas close to the settlement. Settlements are usually not very large, as they are occupied by small groups. Their occupations are short and are characterized by an intensive exploitation of a great diversity of resources. In the archaeological record, this tendency can be observed through a high taxonomic diversity and high anatomical integrity of the faunal remains recovered at the sites since, on the one hand, they do not specialise in a specific taxon and, on the other, because the processing of resources is usually carried out in the settlement itself. The collector model is found in areas where there is less demographic dispersion. These are groups that set up camps that are longer and further apart than those of the foragers. Consequently, the number of settlements throughout the year is smaller. In contrast, the catchment areas are much larger and specialise in a particular taxon. These groups often send out hunting parties in search of herds of a particular species which may undertake expeditions several days away from the settlement. Settlements tend to be large, as they are occupied by larger groups. In the archaeological record, this tendency can be observed through low taxonomic diversity due to hunting specialisation, and low anatomical integrity, as having to transport meat over long distances to reach the settlement means that they make a prior selection of the meat pieces with the highest calorific value at the hunting site.

In the Mediterranean basin of the Iberian Peninsula, during the Epipaleolithic, hunter-gatherer groups are known to have exploited a wide variety of habitats on a seasonal basis (Aura et al. 1998; Aura et al. 2002). The settlements were located at short distances, forming a structured system that combined the exploitation of the coastal lowlands with the inland mountains (Aura and Pérez 1992; Aura et al. 1998, 2002). Most of the settlements are located in rock shelters and caves in a strip no more than 50 km from the present-day coastline. The hunter-gatherer groups were well adapted to this topographical strip of the north-eastern part of the Iberian Peninsula where little distance separates the coast from the inland mountains and with access to a great ecological and bioclimatic diversity without moving long distances and doing so by following the water courses (Aura and Pérez 1992, 1995; Aura et al. 1998).

During the Upper Palaeolithic period, the hunting of horses and aurochs declined, while the mountain goat and red deer became increasingly dominant among the game. In the Epipalaeolithic period, this now well-established dominance is accompanied by the diversification of preys including medium-sized species such as chamois and roe deer (Aura and Pérez 1992; Villaverde and Martínez 1995; Aura et al. 1998, 2002; Villaverde et al. 2012), and most significantly, small game species (Aura and Pérez 1995). The diversification of hunted species has been linked to a reduction in the size of the hunting territories (Aura et al. 1998). Among the small game species, the rabbit was an important one, which began to be hunted in large quantities as a novelty provided by the new ecosystems of the Preboreal and Boreal. Hunting rabbit was intended to complement a diet based on large ungulates (Aura et al. 1998, 2002; Rosado-Méndez et al. 2018, 2019; Villaverde et al. 1998). The remainder of their diet consisted of the hunting of birds and small mammals and collecting wild fruits (Villaverde and Martínez 1995). This diversification of prey and gathering was favoured by the climatic improvement at the end of the Late Tardiglacial and the beginning of the Holocene, which led to an increase in forested habitats and therefore a change in the regional fauna (Villaverde and Martínez 1995). Moreover, from the end of the Late Tardiglacial, the importance of the coastal areas increased, where marine malacofauna and fish were obtained (Aura and Pérez 1995; Villaverde and Martínez 1995; Lloveras et al. 2019; Román et al. 2022). The occupations of these groups were characterized by a preference for ecotones, i.e. areas where they could make a combined exploitation of diverse habitats, which gave them access to a great variety of different resources within a short distance (Aura and Pérez 1995). The coast and inland areas were exploited by taking advantage of the seasonal variations that allowed them to combine gathering molluscs and fishing with hunting large ungulates at certain times of the year (Lloveras et al. 2019; Cuenca-Solana et al. 2021). This dynamic would be linked to a loss of residential mobility and an increase in logistic mobility (Binford 1980; Aura and Pérez 1995; Villaverde et al. 1998).

The aim of this work was to provide information about the mobility of hunter-gatherer groups from the north-east of the Iberian Peninsula during the Epipaleolithic, more specifically, those from Balma del Gai (Moià, Barcelona, Spain). The purpose was to determine what type of occupation (short-, medium-, or long-term) they had and, consequently, to estimate the mobility model of these groups (i.e. forager or collector) especially considering movement between the coastal and inland areas. To do this, a study of dental meso- and microwear of the ungulates hunted by the groups that occupied the Balma del Gai during the Epipaleolithic was carried out, with two main objectives. First, we aimed to determine the diet of the hunted ungulates to reconstruct the habitat in which they lived in the areas near the site. Second, we aimed to study the variability in the diet of the hunted ungulates to determine the duration of the occupations and therefore discern between foragers and collectors regarding the mobility model that hunter-gatherer groups of Balma del Gai had during the Epipaleolithic.

Balma del Gai and archaeological context

Balma del Gai is an archaeological site located in the north-east of the Iberian Peninsula in the municipality of Moià (Barcelona) with the coordinates 2° 08′ 19.5″ E, 41° 49′ 00″ N (Fig. 1A). It is a small rock shelter 10.5 m long and 5.5 m wide, oriented north-east to south-west and with the opening facing south-east (Fig. 1B). The site is at 760 m a.s.l. and is located 50 km from the current coastline. Geographically, it is located on the Moianès plateau, which is on the south-eastern edge of the Ebro depression. It is a region bounded to the east and south by the Bertí crags and to the north by a chain of hills that reach an altitude of 1000 m and progressively lose altitude towards the west, in the direction of the Bages depression and the valley of the Llobregat River.

Fig. 1
figure 1

A Location of Balma del Gai. B Planimetry of the archaeological intervention at the site. C General view of the rock shelter. D Stratigraphic development established through archaeological intervention at the site

The site was excavated on a surface of 20 m2 (Fig. 1B). The stratigraphic sequence contains three levels (Fig. 1D): (1) the surface level, formed by half a meter that contains remains of historic occupations of the shelter and lower-level percolations; (2) Level I, which comprises all the Epipaleolithic occupations; (3) Level II, which consists of gelifracts and is archaeologically sterile (Bergadà 1998). At Level I, two chronological-cultural phases were identified based on the materials recovered and radiocarbon dating. The lower layer, containing Epimagdalenian or Microlaminar lithic complex artefacts, is dated to 13,480–13,233 cal years BP (OxA-29608: 11,440 ± 50 years BP). The upper layer contains the lithic industry of the Filador or Sauveterroid complex dated at 10,368–9,552 cal years BP (Gif-10028: 8,930 ± 140 years BP) (Cuenca-Solana et al. 2021). The dating of the Balma del Gai is consistent with the cultural evolution of the region during the Epipaleolithic (García-Argüelles et al. 2013; Nadal et al. 2017). The Epimagdalenian layer contains a Microlaminar technocomplex with scrapers, points, and bladlets. The upper layer is considered to contain a Geometric or Sauveterroid technocomplex and, to the underlaying layer, includes triangular microliths using the microburin technique. The stratigraphic and lithic homogeneity throughout Level I of Balma del Gai allows us to study this level as a single assemblage and to place it in the Epipaleolithic occupations of the North-eastern Peninsula in a transition period between the Pleistocene and the Holocene (García-Argüelles et al. 2012; Lloveras et al. 2020).

In the zooarchaeological record of the Balma del Gai, the most represented mammal species is the rabbit (Oryctolagus cuniculus), representing 95% of the total individuals identified (Rosado-Méndez et al. 2019). Among the large mammals, the carnivores have a high taxonomic diversity (Nadal et al. 2017) including the Iberian lynx (Lynx pardinnus), the wildcat (Felis silvestris), the fox (Vulpes vulpes), and possibly a wolf (Canis sp.) (Rosado-Méndez et al. 2019). These animals were hunted for their skin using non-discriminatory hunting methods such as traps, but also occasionally for meat consumption, although this would be sporadic and complementary (Nadal et al. 2017). Among the ungulates from Balma del Gai, the most represented is the red deer (Cervus elaphus), which represents 65.4% of the sample. The assemblage also includes small bovids (chamois Rupicapra pyrenaica and ibex or wild goat Capra pyrenaica), large bovids (aurochs Bos primigenius), and wild boar (Sus scrofa). A high percentage of the ungulates displayed pieces of evidence of butchering (Volart-Vázquez 2013).

The study of the raw materials that served as a support for the lithic tools determined that 80% correspond to flint, 15% to quartz and other minority raw materials such as limestone (4%) and some singular supports made on quartz crystal or a single piece made with jasper (Nadal et al. 2017). The extracted limestone is of local origin. The quartz comes from very nearby areas, less than 11 km ± 1 km as well as from some specific places in the municipality of Moià itself (Nadal et al. 2017). Practically all the siliceous materials come from the area of the Congost Valley and Montseny, i.e. at about 30 km ± 3 km (Mangado et al. 2006; Nadal et al. 2017). Furthermore, it was possible to determine that the lithic resources were transported without modification from the procurement areas as all the stages of the production sequence are present at the site (Mangado et al. 2006; Rosado-Méndez et al. 2019). The study of the catchment areas suggests that the hunter-gatherer groups that inhabited Balma del Gai had direct contact with areas close to the coast.

The obtained record of marine malacofauna is also significant, with a total of 43 elements recovered (Estrada et al. 2010), and provides evidence of the relationship of the human groups of Balma del Gai with the coastal areas (Lloveras et al. 2019). The taxonomic diversity stands out if compared with that of other sites in the region of the same chronology, such as the Abric Filador or the Parco Cave (Estrada et al. 2010; Lloveras et al. 2019). The diversity is similar to that found in contemporary coastal sites. This singularity suggests that the groups that occupied Balma del Gai at certain times of the year also occupied regions of the central Catalan coast where they had direct access to these resources, unlike other inland settlements where they obtained them through exchange (Estrada et al. 2010; Nadal et al. 2017). These pieces of evidence of groups’ mobility with an alternation between the coast and the interior for the use of resources by the hunter-gatherer groups that inhabited the Balma del Gai raise the hypothesis of possible seasonal occupation of the site that needs to be tested.

Materials and methods

Sample selection

The materials analysed in this work were obtained from the 1994–2016 excavations of Balma del Gai. The material is housed at the Laboratory of Archaeology of the Department of History and Archaeology of the University of Barcelona.

All dental remains of ungulates from Level I were reviewed. Among all teeth, we selected the fourth premolars (P4/p4) and the first, second, and third molars (M1/m1, M2/m2, and M3/m3), both upper and lower, according to the protocol established by Rivals et al. (2015) and Xafis et al. (2017). Teeth belonging to the youngest individuals that do not present wear facets were discarded (Rivals et al. 2007) as well as those that had a damaged occlusal surface due to taphonomic processes (King et al. 1999; Uzunidis et al. 2021; Micó et al. 2023).

Following these criteria, 31 dental remains were selected, each one corresponding to different individuals and belonging to three different species: Cervus elaphus (N = 8), Rupicapra pyrenaica (N = 13), and Sus scrofa (N = 10).

Mesowear

The mesowear analysis, developed by Fortelius and Solounias (2000) to infer diet and habitat of herbivorous mammals, consists of macroscopic observation of the morphology of the dental cusps in ungulates, ideally the paracone of upper molars, although it has also been successfully performed on other tooth positions (Kaiser and Solounias 2003), as well as on the lingual cusps of lower molars, i.e. with paraconid and metaconid (Kaiser and Fortelius 2003; Fraser et al. 2014). The method evaluates the relief and sharpness of the cusps to relate it to the degree of attrition and abrasion of the diet. Browsers have a relatively soft diet as they consume dicotyledonous plants and present high and sharp cusps. On the other hand, grazers that consume monocotyledonous plants, especially grasses with a higher concentration of phytoliths, present low and rounded/blunt cusps. Mesowear reflects the diet of an animal during the months or years before its death (Kaiser and Schulz 2006; Rivals et al. 2007; Louys et al. 2012; Uzunidis et al. 2023), although if the diet is extremely abrasive, the wear of the cusps could occur more quickly (Solounias et al. 2014). However, mesowear does not detect dietary changes in shorter periods of time, such as the seasonal changes in food that can occur in a population (Rivals et al. 2013).

The mesowear analysis was conducted on Cervidae and Bovidae. The Suidae were excluded because mesowear is a method developed for selenodont and lophodont teeth, not for bunodont teeth such as those of the suids (Fortelius and Solounias 2000; Green and Croft 2018).

In this study, the standardized method introduced by Mihlbachler et al. (2011) and modified for cervids and bovids by Rivals et al. (2017) was employed. This method is based on a mesowear “ruler” that has seven categories (from 0 to 6), ranging from high and sharp cusps (category 0) to low and blunt cusps (category 6). Mesowear was analysed from the buccal view of the upper teeth and the lingual part of the lower teeth (Solounias and Semprebon 2002). The degree of wear was noted and averaged for each taxon to obtain the mesowear score (MWS).

Microwear

Microwear analyses detect microscopic features left by abrasive particles on the enamel of the occlusal surface during chewing. The constant formation of new features erases the previous ones, leaving a record of the diet on a much shorter time scale compared to mesowear (Grine 1986). Therefore, microwear analysis studies the diet from days, weeks, or even hours before the death of the animal (Solounias and Semprebon 2002).

The samples were prepared following the protocol developed by Solounias and Semprebon (2002) and Semprebon et al. (2004). To remove residues of consolidants, dust, or sediment, the occlusal surface was cleaned using acetone then 96% ethanol. A high-precision mould of the surface was made using polyvinylsiloxane silicone (Heraeus Provil novo, Light CD2, regular set) and a second layer using silicone putty (Heraeus Provil novo, Putty). Finally, transparent casts were produced using high-resolution epoxy resin (CTS EPO150 resin + K151 hardener).

The casts were observed using a standard light stereomicroscope (Zeiss Stemi 2000C) at × 35 magnification to identify and quantify the microfeatures on the enamel bands of the occlusal relief and were quantified in an area of 0.16 mm2 using an optical reticule (Solounias and Semprebon 2002; Semprebon et al. 2004). Quantification was done on the mesial area (paracone or protocone) of the upper teeth and on the mesiolingual area (protoconid) of the lower teeth as established by Solounias and Semprebon (2002). All specimens were analysed by a single observer (PDI) to avoid an inter-observer error. Photomicrographs were taken using an Invenio 5SII digital camera and DeltaPix InSight software in extended focus mode which merges images acquired at various focal planes to produce a single image with a greater depth of field.

The microscopic marks can be classified as pits and scratches, although there are several types of them (Fig. 2) (Solounias and Semprebon 2002; Semprebon et al. 2004). Pits have a circular or semi-circular shape of similar length and width and are classified as either small or large. Small pits are characterized by being shallow and appear as shiny, bright, white dots. Large pits are wider and deeper and therefore reflect less light. Puncture pits are very deep and symmetrical, with a crater-like morphology with regular margins, and appear as dark dots. Gouges are crater-shaped, but unlike the puncture pits, have irregular and asymmetrical margins. Scratches are elongated and narrow features. They are classified as fine, coarse, or hypercoarse scratches. Fine scratches are narrow and very shallow, while coarse scratches are wider and deeper. Both fine and coarse scratches reflect light, appearing bright under the microscope. Hypercoarse scratches are very deep and are wider than fine and coarse scratches, and consequently they have very low reflectivity and appear as dark features.

Fig. 2
figure 2

Photomicrograph of the tooth enamel surface (× 35 magnification) of an individual of Cervus elaphus from Balma del Gai (Site Inventory #7). Scale bar = 0.4 mm. Features: fine scratches (FS), coarse scratches (CS), hypercoarse scratch (HS), small pit (SP), large pit (LP), and gouges (G)

In this study, the following variables were used: (1) total number of scratches, (2) total number of pits. For these two variables, two counts were made in two different areas of the occlusal surface and averaged to obtain a mean per individual. (3) Presence of more than four large pits. (4) Presence of more than four gouges. (5) Scratch width score (SWS), which was obtained by giving a score of 0 to teeth with predominantly fine scratches, a score of 1 to teeth with a mixture of coarse and fine scratches, and a score of 2 to teeth with predominantly coarse scratches. (6) Presence of more than four cross scratches.

The main dietary categories of a population are distinguished by computing the percentage of individuals per taxon with a low number of scratches (0–17), which indicates the different dietary strategies in a population (Solounias and Semprebon 2002). Browsers display a low scratch tendency with a range between 72.73 and 100%. On the contrary, grazers show lower percentages fluctuating from 0 to 22.2% of individuals falling between 0 and 17 scratches. Mixed feeders overlap with grazers because they fall between 20.93 and 70% of individuals.

Comparison with extant species

The mesowear and microwear results were compared to the dataset of extant ungulates with known diets from Fortelius and Solounias (2000) and Solounias and Semprebon (2002), respectively. Populations of extant ungulates from Rivals et al. (2015) and a new dataset produced by FR were added to compare directly related species.

Red deer (Cervus elaphus) were compared with extant populations of the same species corresponding to four locations: (1) Central Europe (N = 32), (2) Isle of Rum, Scotland (N = 211), (3) Pyrenees (N = 1), and (4) Riaño (León, Spain) on the southern slope of the Cantabrian Range (N = 23).

Pyrenean chamois (R. pyrenaica) were compared with extant populations corresponding to six different species of Caprinae: (1) Capra ibex (N = 18) from the Alps, (2) C. pyrenaica pyrenaica (N = 4) from the Western Pyrenees, (3) C. p. victoriae (N = 7) from the middle-mountain areas of the Spanish Central System, (4) R. p. pyrenaica (N = 1) from the Pyrenees mountains, (5) R. rupicapra. caucasica (N = 3) that inhabits the Caucasian Range, and (6) R. r. rupicapra (N = 19), from central European alpine areas.

Wild boars (Sus scrofa) were compared with extant populations corresponding to three locations: (1) Central Europe (N = 15), (2) Israel (N = 31), and (3) Tunisia (N = 2).

To compare the numbers of pits and scratches between different taxa, an analysis of variance was applied (One-way ANOVA) with a post hoc Tukey HSD.

Seasonality

Finally, the method developed by Rivals et al. (2015) based on the variability of scratches, i.e. the standard deviation (SD) and the coefficient of variation (CV), was applied to estimate the duration of the accumulation of the faunal assemblages. The method relies on the seasonal variability of the diet; if the hunted animals died rapidly or at the same time of the year, the variability will be low, while if they died over a year or more, there will be higher variability. Combining the two values (SD and CV) on a bivariate plot allows the classification of the samples into three categories: (a) seasonal occupation or shorter, (b) occupation longer than a season (one or more contiguous seasons), and (c) separated events that took place in non-contiguous seasons. Only the red deer and the Pyrenean chamois were selected for this method because it was solely developed for pure herbivores.

Results

The dietary traits of the ungulates C. elaphus, R. pyrenaica, and S. scrofa from Balma del Gai were studied through dental mesowear and microwear analyses. The results are summarised by species in Table 1. Raw data from all the specimens sampled are available at https://doi.org/10.5281/zenodo.7377039 (Duñó-Iglesias 2022).

Table 1 Summary of mesowear and microwear data from the fossil assemblage of Balma del Gai. Abbreviations: N, number of specimens; MWS, mesowear score; NP, average number of pits; NS, average number of scratches; %LP, percentage of individuals with large pits (> 4); %G, percentage of individuals with gouges (> 4); SWS, scratch width score; %XS, percentage of individuals with cross scratches (> 4); M, average; SD, standard deviation; CV, coefficient of variation

Mesowear

From a total of 31 individuals, nine individuals were suitable for the mesowear analysis, and 22 had to be discarded due to the poor condition of the cusps that had been altered by taphonomic processes or had not yet sufficiently developed their wear facets due to ontogenetic causes (Rivals et al. 2007).

Results for the red deer were obtained from four individuals. The mesowear pattern tended to display relatively high and sharp cusps, ranging from 1 to 3 on the mesowear “ruler” with an average (MWS) of 2, which indicated a medium abrasion diet, corresponding to a mixed feeder signal.

Results for the chamois were obtained from five individuals. The mesowear pattern tended to display high and sharp cusps ranging from 1 to 3 on the mesowear “ruler” with an average (MWS) of 1.6, which indicated a medium abrasion diet, corresponding to a mixed feeder signal.

Comparison of the mesowear of the ungulates from Balma del Gai with the extant species

The red deer from Balma del Gai (MWS = 2) was compared with extant red deer from different locations. Its MWS is similar to that of C. elaphus from Central Europe (MWS = 1.67) and, as can be seen in the box diagram (Fig. 3), it falls within its range. Furthermore, although there is only data of one individual of C. elaphus from the Pyrenees with a MWS of 2, it coincides with the degree of abrasiveness of the diet of C. elaphus from Balma del Gai. Although statistically insignificant, it provides an approximate value of the abrasiveness of the diet.

Fig. 3
figure 3

Box plot featuring mesowear scores (MWS) of the extant species from different locations: extant red deer in green; extant wild goats in yellow; extant chamois in purple. We also display the feeding regimes (grazer, leaf browser, mixed feeder) of extant ungulates from Fortelius and Solounias (2000). All were compared to the values of the mesowear score (MWS) from the archaeological species from Balma del Gai, C. elaphus and R. pyrenaica in red

The Pyrenean chamois from Balma del Gai, despite having a small number of individuals (N = 5), provides an approximate value of the abrasiveness of its diet (MWS = 1.6). This value is similar to that of the current species of the genus Rupicapra, as can be seen in the box diagram. The species R. r. rupicapra, which inhabits the regions of central European alpine areas, has a mean MWS value of 1.3 (N = 30), which is very close to that obtained for the R. pyrenaica from Balma del Gai. The difference in the magnitude of the samples could explain the small divergence between the two values.

In extant R. r. caucasica, despite having a statistically non-significant number of individuals (N = 4), the abrasiveness of its diet is quite close to that of extant R. r. rupicapra and that of R. pyrenaica of Balma del Gai, with a MWS of 1.25. Rupicapra pyrenaica pyrenaica is only represented in the extant sample with a single individual that has a MWS of 0. The value is not significant; however, it corresponds to a low abrasion diet, i.e. a browsing diet, quite far from the tendency shown by R. pyrenaica from Balma del Gai (MWS = 1.6).

In general, current Capra tend to display diets of very low abrasion, with values close to 0.8. On the other hand, C. p. pyrenaica (collected in the twentieth century, now extinct) presents a very high MWS of 2.33, which is situated in a medium–high abrasion diet. The singularity is that because we had a very low sample of only three individuals, a single extreme value (out of three individuals, two had a MWS of 1 and one had a MWS of 5) makes the average increase considerably. Nevertheless, the sample is too small to conduct any statistical analysis. These data indicate that the extant species of the Capra genus would have had a diet with a tendency towards browsing, rather than R. pyrenaica from the Balma del Gai, which tended to have a mixed diet throughout the year. However, it should be noted that the sample is statistically very small, which limits the reliability of its interpretation.

Microwear

For the microwear analysis, a total of 18 individuals out of the whole sample gave positive results when observed under the stereomicroscope. Therefore, 13 dental remains had to be discarded because of the poor state of the wear facets or the absence of these due to taphonomic or ontogenetic causes. The 18 individuals analysed are representative of the three species studied.

For the red deer, results were obtained from seven individuals. The microwear pattern was characterized by a high number of pits, ranging from 11.5 to 38.5 and averaging a mean of 27.5 pits per counting area, and a relatively low number of scratches, ranging from 11 to 15.5 and averaging a mean of 12.78 scratches. A flat occlusal surface with plucked prisms was observed (Fig. 4B), typical features of an attritive diet. The presence of large pits (LP) was observed in 57% of the individuals. Also, 57% of individuals observed displayed gouges (G). A scratch width score (SWS) with a mean of 1.14 indicated a mixture of fine and coarse scratches in equal proportions. However, it was observed that some individuals showed a higher proportion of coarse than fine scratches. Finally, no cross scratches (XS) were observed.

Fig. 4
figure 4

Photomicrographs of selected fossil ungulate tooth enamel surfaces (× 35 magnification) from Balma del Gai. Scale bar = 0.4 mm. A Cervus elaphus (Site Inventory #7) with fine scratches but also coarse and hypercoarse. Also, pits of different sizes and depths can be observed; small pits, large pits, and gouges. B Cervus elaphus (Site Inventory #48) with scratches of different sizes and depths, corresponding to both fine and coarse scratches and with a high number of pits, although those are harder to distinguish. C Sus scrofa (Site Inventory #373) with many fine scratches and some coarse ones. Also, the presence of cross scratches was detected. Many small pits, some large pits, and some gouges can be observed as well

Results for the Pyrenean chamois were obtained from three individuals. The microwear pattern was characterized by a high number of pits, ranging from 21 to 24.5, with a mean of 21.66 pits per counting area and a relatively low number of scratches ranging from 11 to 15, averaging a mean of 13.16 scratches. The presence of large pits (LP) and gouges (G) was observed in all three individuals analysed. The scratch width score (SWS = 1) indicated a mixture of scratches of different sizes, with an equal proportion of fine and coarse scratches. All scratches had similar orientations and no cross scratches (XS) were observed. Although Rupicapra pyrenaica is only represented by three individuals, it provides an approximate tendency of its dietary traits.

Results for the wild boar were obtained from 8 individuals. The tooth microwear pattern was characterized by a high number of pits, ranging from 14 to 45, averaging a mean of 21.83 pits, and a low number of scratches, ranging from 10 to 16.5 averaging a mean of 13.81 scratches per counting area. The presence of large pits (LP) was observed in 57% of the individuals and gouges in 37%. The presence of cross scratches was observed in 50% of the studied individuals.

All the individuals of the archaeological sample showed a low number of scratches with less than 17 scratches per counting area (i.e. high %0–17) for all the individuals of the three taxonomic populations analysed. This suggests that all the individuals analysed had browsing dietary traits at the time of their death.

The three species from Balma del Gai are found in the browsing ecospace (Fig. 5). The red deer and the Pyrenean chamois fall within the ellipse corresponding to the browsing diets, and their error bars (corresponding to the standard deviation) are maintained within this region. The wild boar also fall within the ellipse of the browsers, although its error bars indicate a certain variation that includes an occasional mixed feeder diet and a greater variability in comparison to other species of Balma del Gai. To summarise, the microwear analysis of the three samples clearly indicates that all the species had a browsing diet in the days or weeks prior to their death.

Fig. 5
figure 5

Bivariate plot of the mean number of pits and scratches for the three different species of ungulates from Balma del Gai. The error bars correspond to the standard deviation (SD). The ellipses correspond to Gaussian confidence ellipses (p = 0.95) for extant ungulate browsers and grazers (Solounias and Semprebon 2002)

Comparison of the microwear of ungulates from Balma del Gai with extant species

Cervus elaphus

The red deer from Balma del Gai was found to have a similar diet to that of extant browsers, together with the other red deer populations that were used for comparison (Fig. 6). Cervus elaphus from Balma del Gai was found within the same scratch range as extant C. elaphus populations (Table 2; Fig. 5). On the other hand, the mean number of pits is notably higher in the case of Balma del Gai when compared to populations from Riaño and Central Europe which have fewer pits (Table 2; Fig. 6). However, it should be noted that in the case of the population from the Isle of Rum which has the highest number of individuals (N = 249) and, consequently, the most reliable number statistically speaking, it is considerably closer to the average number of pits of Balma del Gai (Table 2; Fig. 6). Finally, the individual from the Pyrenees displays an extremely high number of pits (Npits = 44.5) showing a specific dietary trend for an area not that far from that of Balma del Gai (Table 2).

Fig. 6
figure 6

Bivariate plot of the mean number of pits and scratches for C. elaphus from Balma del Gai in comparison with extant populations from Central Europe, Isle of Rum, Riaño, and the French Pyrenees. The error bars correspond to the standard deviation (SD). The ellipses correspond to Gaussian confidence ellipses (p = 0.95) for extant ungulate browsers and grazers (Solounias and Semprebon 2002)

Table 2 Microwear data of C. elaphus from Balma del Gai and extant populations from Central Europe, Isle of Rum, Riaño, and the French Pyrenees. Abbreviations: N, number of specimens; NP, number of pits; NS, number of scratches; %LP, percentage of individuals with large pits (> 4); %G, percentage of individuals with gouges (> 4); SWS, scratch width score; %XS, percentage of individuals with cross scratches (> 4); M, average; SD, standard deviation; CV, coefficient of variation

Rupicapra pyrenaica

All extant Caprinae (genus Capra and Rupicapra) are found in the region of the graph corresponding to mixed feeding diets, except for C. ibex, which remains in the grazing region, although at the limit of it, with a certain proximity to the mixed diet area (Fig. 7). On the other hand, R. pyrenaica of Balma del Gai have a clear tendency for browsing, which would stand out for the very low number of scratches compared to the extant species of Caprinae (Fig. 7; Table 3). It should also be noted that the number of pits (Table 3) remains within the range of the current Rupicapra (Fig. 7).

Fig. 7
figure 7

Bivariate plot of the mean number of pits and scratches for R. pyrenaica from Balma del Gai compared to extant Caprinae; R. p. pyrenaica, R. p. caucasica, R. r. rupicapra, C. p. pyrenaica, C. p. victoriae and C. ibex. The error bars correspond to the standard deviation (SD). The ellipses correspond to Gaussian confidence ellipses (p = 0.95) for extant ungulate browsers and grazers (Solounias and Semprebon 2002)

Table 3 Microwear data from the Caprinae: R. pyrenaica from Balma del Gai and extant populations of R. p. pyrenaica, R. p. caucasica, R. r. rupicapra, C. p. pyrenaica, C. p. victoriae, and C. ibex. Abbreviations: N, number of specimens; NP, average number of pits; NS, average number of scratches; %LP, percentage of individuals with large pits (> 4); %G, percentage of individuals with gouges (> 4); SWS, scratch width score; %XS, percentage of individuals with cross scratches (> 4); M, average; SD, standard deviation; CV, coefficient of variation

Sus scrofa

The wild boar from Balma del Gai shows a trend that is different from the current S. scrofa, which tends to have a mixed diet (Fig. 8). This trend is characterized by a low number of scratches, as in C. elaphus and R. pyrenaica (Table 4; Fig. 8). It should be noted that the number of pits is high, but quite similar to that of S. scrofa from Central Europe and Tunisia (Table 4; Fig. 8).

Fig. 8
figure 8

Bivariate plot of the mean number of pits and scratches from the S. scrofa from Balma del Gai compared with the extant populations from Tunisia, Central Europe, and Israel. The error bars correspond to the standard deviation (SD). The ellipses correspond to the Gaussian confidence ellipses (p = 0.95) for extant ungulate browsers and grazers (Solounias and Semprebon 2002)

Table 4 Microwear data of Sus scrofa from Balma del Gai and extant populations from Tunisia, Israel, and Central Europe. Abbreviations: N, number of specimens; NP, average number of pits; NS, average number of scratches; %LP, percentage of individuals with large pits (> 4); %G, percentage of individuals with gouges (> 4); SWS, scratch width score; %XS, percentage of individuals with cross scratches (> 4); M, average; SD, standard deviation; CV, coefficient of variation

Seasonality

Using the data obtained from the microwear analysis and applying the method developed by Rivals et al. (2015), it was possible to determine the distributions of mortality for species from Balma del Gai (Fig. 9). The individuals of the populations of red deer and Pyrenean chamois had a very low variability in the number of scratches (in SD and CV) and fall in region A of the plot (Fig. 8). This indicates that they died during a period equal to or less than a single season.

Fig. 9
figure 9

Boundary lines separating three regions with the probability of error (heat map) based on the values of the SD (standard deviation) and the CV (coefficient of variation), and which serve to classify the populations studied according to the distribution of mortality during the year. They correspond to a seasonal mortality (equal to or less than a season) [A], a mortality throughout the year (more than one season) [B], or mortality events at separate times of the year (in different non-contiguous seasons of the year, for example, spring and autumn) [C]. This graph corresponds to the ungulates of Balma del Gai

Discussion

Dietary traits and habits of the ungulates from Balma del Gai

Red deer from Balma del Gai have mesowear that reflects mixed feeding dietary traits, i.e. this population would alternate between browsing and grazing on an annual basis. Throughout the year, red deer had access to an open, grassland-type habitat, but also dwelled in woodland-type habitats with a more closed and wooded structure. As far as microwear is concerned, red deer had a pure browsing diet indicated by a low number of scratches and presence of plucked prisms. When it is compared with present-day populations, it is very similar to the C. elaphus that inhabit the island of Rum. This does not mean that they fed on the same plants, but that the structure of the vegetation was similar. According to the comparison of archaeological and extant red deer, the structure of the vegetation in the surroundings of Balma del Gai would be similar to that found on the island of Rum today, mainly grassy and with small bushes (Gordon and Illius 1989).

Regarding the Pyrenean chamois, mesowear indicates a mixed feeding diet. On an annual scale, the Pyrenean chamois would alternate between browsing and grazing diets, i.e. seasonally shifting between open and wooded habitats. As far as microwear is concerned, it has a clear browsing character much more marked than in extant Caprinae of the genus Capra and Rupicapra. More specifically, focusing on extant chamois, microwear indicates a mixed feeding diet, which agrees with the fact that the current chamois are typical mid-high mountain ungulates, and their altitudinal range in the Pyrenees usually varies between 1000 and 2800 m (Corlatti et al. 2022). Evidence from the fossil record of Pleistocene and Holocene sites shows that in the past, the chamois was present in low-lying areas close to the sea throughout the northern part of the Mediterranean basin (Rivals et al. 2005). For this reason, the diet of the chamois found at Balma del Gai differs from that of modern-day chamois or wild goats. This evidence provides insight into the intrinsic habitat of the chamois and its dietary adaptations to a wider range of altitudes and habitats, including pre-coastal and coastal areas, in addition to the mid-to-high mountain ranges where they are found today. Furthermore, this result exemplifies the chamois’ dietary adaptability, enabling it to inhabit a wide range of habitats. Anthropic pressure would be a factor that would drive them to take refuge at higher altitudes, constraining its dietary flexible capabilities to the mixed feeding dominated niche. The presence of chamois in Epipaleolithic chronologies suggests that there was minimal anthropogenic pressure in the north-east of the Iberian Peninsula during that period. For the Iberian ibex (Capra pyrenaica), a similar species in terms of taxonomy and ecology, the geographic distribution area reaches its peak during the second half of the Upper Pleistocene spreading from the coastal and pre-coastal areas to altitudes below 1000 m in the Pyrenees (Fosse et al. 2021). During the Postglacial period, it is still present from the Cantabrian range to the eastern Pyrenees. During the Epipalaeolithic, in the south-eastern slope of the Pyrenees, it is already limited to interior sites and always at an altitude corresponding to the mid-mountain zone. From the Neolithic to the Bronze Age, its distribution progressively becomes limited to the higher areas of the Pyrenees (Fosse et al. 2021). It is present until the end of the twentieth century on the southern slope of the central Pyrenees. Hence, it is reasonable to assume that the displacement of the chamois ecological niche occurred during later periods of the Holocene.

The mesowear method is not suitable for the wild boar due to its bunodont teeth. The microwear pattern observed on the wild boar from Balma del Gai, characterized by a low number of scratches and a high number of pits, corresponds to a browsing diet. The extant wild boar consumes both plants and roots and tubercles, and on rare occasions, it also has a scavenger diet (Scandura et al. 2022), which is consistent with the abundance of puncture pits and cross scratches identified on the wild boar from Balma del Gai that would indicate the consumption of fruits/seeds, and support multidirectional jaw movements, respectively.

The results obtained in this study indicate that the structure of the vegetation during the Epipaleolithic in the surrounding areas of Balma del Gai was a landscape that would have alternated open habitats with grasses with habitats comprising thickets of low shrubs. On the other hand, the results obtained for the chamois display the intrinsic dietary adaptations of this species in its lost habitats at low altitudes. The confinement of extant chamois in mountainous areas is due to a latter anthropic pressure.

Seasonality and duration of occupation

For both the red deer and the Pyrenean chamois, different results were obtained through dental mesowear and microwear analyses (Table 5). For the two species, mesowear indicates a mixed feeder diet (MF) and microwear suggests a browsing diet (B). The discrepancy between the results of mesowear and microwear is due to the different time scales recorded with each method. The mesowear provides information on the average diet of a population over the course of the year, while the microwear indicates the diet at a given time, just before death, i.e. it records a strong seasonal signal. Thus, the discrepancy between the results of the mesowear and microwear supports high seasonal shifts in diet and due to the temporal resolution of the microwear signal, a browsing diet at the time of death of these individuals (Sánchez-Hernández et al. 2016). Consequently, it suggests that there is also a high seasonality in the occupation patterns of the site. The red deer and Pyrenean chamois were mixed feeders but died at a season when they were browsing.

Table 5 Synthesis of the results obtained for mesowear and microwear, as well as the seasonality of the distribution of mortality of the individuals from Balma del Gai. For both C. elaphus and R. pyrenaica, the mesowear column indicates a mixed feeder diet (MF), and the microwear column indicates a browser diet (B). On the other hand, through seasonality methods, a mortality equal to a season or less was obtained for both the Pyrenean chamois and the red deer

The analysis of the variability in the microwear scratches provides support for the seasonality of accumulation of the ungulates at the site.

For both the red deer and the Pyrenean chamois, the low variability of the scratch density suggests that the animals were hunted during short seasonal periods, covering a time period of one season or less. This indicates that these two species were hunted during a very short and specific period of the year.

Mesowear classified the feeding habits of the red deer from Balma del Gai as a mixed diet, i.e. varying seasonally between grazing and browsing. Microwear indicates that red deer had a browsing diet in the days before death. In the Iberian Peninsula, modern-day red deer feed on a wide variety of plant species, both grasses (Gramineae) and ligneous plants. The diet is mainly composed of herbaceous plants in winter and especially in spring. In summer and autumn, there is a gradual increase in the proportion of woody plants (tree leaves and bushes) consumed, which corresponds to the seasonal use of these habitats (Braza et al. 1984; Braza and Álvarez 1989; Escós and Alados 1992; José et al. 1997; Carvalho et al. 2012; Alves et al. 2014).

In summary, red deer have a grazing diet in winter and spring and shift to browsing in summer and autumn. This would indicate that the red deer from Balma del Gai, which were browsing at the time of death, would have been hunted during the late summer to autumn period. Furthermore, this hypothesis is supported by other evidence recovered at Balma del Gai. First, the abundance of grove snails (Capaea nemoralis) indicates that the occupation of the rock shelter took place in periods that were neither intensely cold nor intensely hot, i.e. during spring to early summer or late summer to autumn. Two other pieces of evidence allow us to narrow down the seasonality, namely the consumption of blackthorn berries (Prunus spinosa) that are available from the end of summer but are edible only after the first frost of the autumn together with the identification of red deer antlers that were not shed and were well developed. This suggests that animals were hunted when the antlers were in the last stages of formation or were already fully formed, which would correspond to autumn (Lloveras et al. 2011: 43).

The studies on seasonality and variability indicate that the occupation of Balma del Gai took place during a very short and specific period of the year. On the other hand, the diet of the deer, together with other evidence found at the site, indicates that these short seasonal occupations of the Balma del Gai site would have taken place in the late summer and especially autumn, i.e. during the mating season when ungulate herds like red deer or chamois aggregate to breed, which makes them more vulnerable (Corlatti et al. 2022; Mattioli et al. 2022).

Mobility of the human groups

During the Epipalaeolithic, human groups had a high mobility between different settlements in close proximity to each other, in which they exploited a wide variety of resources in very small catchment areas close to each settlement (Aura and Pérez 1992; Aura et al. 1998, 2002). This study provides new evidence on the settlement model of the human groups at Balma del Gai. The short-occupation settlement at a very specific season of the year, together with the evidence of a high taxonomic diversity and a high anatomical integrity of the faunal remains recovered, supports the hypothesis of a forager-type occupation and mobility model. The groups that occupied Balma del Gai in late summer and autumn would have been in other locations at other times of the year, i.e. winter and spring, possibly close to the coast, as suggested by the evidence of the marine malacofauna (Lloveras et al. 2019), as well some raw material (jasper) obtained from outcrops that only appear in the coast of Barcelona (Carbonell et al. 1997).

Conclusions

The combined use of tooth meso- and microwear provides new information on the occupation and mobility patterns of hunter-gatherer groups in the north-east of the Iberian Peninsula during the Epipaleolithic. It supports the hypothesis that Balma del Gai was a forager settlement, occupied during late summer and autumn by hunter-gatherer groups that inhabited areas close to the coast at other times of the year. This seasonal alternation between the coast and the inland is very well adapted to the topography of the north-east of the Iberian Peninsula, as there is a short distance between the coast and the mountains of the inland, with access to a high ecological diversity for the exploitation of different resources. Balma del Gai would have been a settlement that was occupied at a very specific time of the year for short periods when specific resources such as ungulates were easy to exploit.

In a more general context, this forager model of mobility found at Balma del Gai could be reproduced at other sites of the Mediterranean coast and pre-coastal area of the north-east of the Iberian Peninsula during the Epipaleolithic period. It is a model that adapts very well to the topography of this region, allowing the maximum benefit to be drawn from both coastal and inland areas that offer different resources at specific times of the year. The mobility pattern is characterized by a seasonal specialisation in different biotopes, through settlements of logistical type and short occupations, where they obtained certain resources. At the same time, however, they exploited a wide range of resources to supplement their diet and obtain raw materials. Furthermore, the occupation of inland sites could take place during late summer and especially autumn, while winter and spring would be spent in areas near the coast, making the most of the seasonal resources offered by the Mediterranean climate. Further microwear studies need to be carried out at other sites chronologically contemporary in the region to confirm this mobility pattern and the seasonal mobility between the interior and the coast in the north-east of the Iberian Peninsula during the Epipaleolithic.