Rock Art and Social Memory in the Deseado Massif: An Approach from the Study of Superimpositions in Cueva 2, Los Toldos, Santa Cruz Province, Argentina

Abstract

In this work, we explore the way in which rock art mediates social memory. Our study case is based on the rock art sequence established for Los Toldos archaeological locality, Argentina, which begins in the Pleistocene/Holocene transition and extends to the Late Holocene. The analysis focuses on superimpositions, with an emphasis on the human attitudes towards previous images. Despite changes detected along the sequence, Los Toldos manifests a strong emphasis on recalling the past, which is evidenced by the superimposition, replication, maintenance and recycling of motifs. These behaviors show a multi-generational dialogue that kept an ancestral memory alive but also recreated it through grouping images from different times for telling stories. This study places an emphasis on the users/consumers of rock art rather than on its makers. This focus narrows the gap between the archaeological record and the ethnographic sources by claiming that the Tehuelche people were engaged with rock art even though they did not make it.

1. Introduction

The objective of this paper is to understand the role of rock art in the mediation of social memory among the hunter gatherer societies who inhabited the Deseado Massif (Santa Cruz province, Argentina) from the late Pleistocene to the final Late Holocene (Figure 1). For this purpose, we analyze the superimposed pictographs of Cueva 2 site (Los Toldos locality) to evaluate human attitudes towards previous images (Re 2016). The concept of “collective memory” was developed by the French sociologist Maurice Halbwachs in the first half of the 20th century, who proposed that memory was made through social interaction. This idea meant that what people remembered was shaped by the social contexts in which they lived, and it provided them with a sense of identity (Hendon 2020). Currently among scholars, “collective memory” generally refers to the creation and maintenance of nationalist identities, while “social memory” describes more broadly the relational and contested ways in which groups construct their identities (Jones and Russell 2012; Van Dyke 2019). According to the latter contributions, images, objects, monuments, human burials, oral histories, stories, folklore, myths, events and places are important mediators of social memory, beyond written sources. Although not exclusive to hunter gatherers, these mnemonic devices have an important significance in the way in which these societies pass memory through commemorative ceremonies, storytelling and bodily practices (Connerton 1989; Munn 1973). The concept of “place”, theoretically developed in the 1990s within anthropology and landscape archaeology, is central to this discussion. According to Ingold’s dwelling perspective (1993), landscape is a testimony to the lives and works of the many generations who have inhabited the world. People build their identities through their actions and movements across the land and through the actions and movements of their ancestors. Thus, places and the pathways that connect them acquire history (Ingold 1986, 1993; Tilley 1994). An example is the importance of the Dreaming (or Dreamtime) among Australian Aboriginal societies. Even though it refers to the deeds of their Ancestors in the beginning (the time when the land was created, and people were placed on it), it is evident in the landscape features with whom people interact in daily life. While the Dreamtime works as a constant aid memoire about their origins, it also reminds them about the location of important places and resources, such as water and ochers (Blundell 1980; Flood 2004). Schlanger (1992) defined “persistent places” as repeatedly occupied spots within the long-term use of a region, due to specific qualities that made them suitable for certain practices or activities. These places evidence redundant and intense occupations (Figuerero Torres and Mengoni Goñalons 2006) and tend to be signaled by special features that have motivated their reoccupation. Some of them are human burials (Martínez et al. 2006; Prates and Di Prado 2013), rock art and stone structures (Carden et al. 2023; Miotti et al. 2016). Persistent places are, thus, especially appropriate for evaluating the social construction of memory and human attitudes towards the past.

There is a vast body of archaeological literature dealing with the active role of rock art in the construction of social memory. Through a cognitive approach, Mithen (1988) interpreted Palaeolithic art as an educational device. Zoomorphic figures displayed special cues that worked as memory aids for apprentices who were learning to track animals. A similar idea was proposed for the guanaco hunting scenes painted in Cueva de las Manos, Argentina (Aschero and Schneier 2023). Petroglyphs and pictographs are one of the few cultural products that are never transformed into an archaeological record because, “…as long as they are in view, they are constantly resignified, playing roles in social life, whether of anecdotal or radical importance” (Armstrong Bruzzone 2012, p. 19). As focal points where the past has persisted into the present, rock art panels have enabled relationships between people who were not in direct contact, linking distant generations (Armstrong Bruzzone 2012; Martel et al. 2012; Quesada and Gheco 2015). Images exposed on rock surfaces could be reused or recycled, with or without modifications, to integrate them into new compositions. They could also act as visual models for future representations. From these properties, Aschero (1996) defined rock art as an “additive phenomenon”. These ideas are related to Eco’s ([1962] 1992) concept of “open work” (Gheco et al. 2013; Quesada and Gheco 2015; Recalde and Colqui 2022; Troncoso Meléndez 2008).

Even though its original meanings tend to be lost with time elapsed and populational discontinuities, rock art’s permanence in the landscape still activates memory. This is why Norder (2012) proposed to focus on the user/caretaker context rather than on the maker/meaning framework. His perspective led other researchers to consider the consumption of rock art through time instead of its creation and original meanings, since images are not inert representations of the world (Creese 2021; Motta 2019; Zawadzka 2021). Another point worth considering is that rock images are just a part of the much wider process of memory making, which also involves singing, dancing, telling stories, listening, smelling and using specific objects. In this sense, rather than being fixed to a rock substrate (as they physically are), these images circulated across the landscape through people’s knowledge and stories (O’Connor et al. 2022; Zawadzka 2021; Zubieta 2022).

The theoretical interest in social memory has brought special attention to the analysis of superimpositions in rock art studies. Beyond constructing relative chronologies, the objective of such studies is understanding human attitudes towards previous images. Some of these identified intentions are breaking with preexisting rock art through destroying or annulling the original images, and making a difference with the past but still respecting it by leaving most of the underlying images visible (Re 2016; Recalde and Colqui 2022; Vergara and Troncoso 2016). Other attitudes include incorporating the older images into new compositions (Aschero and Schneier 2023; Gheco et al. 2013; Martel et al. 2012; Motta 2019), generating a ritual sensation of primordial and untouched places through minimal and almost imperceptible superimpositions (Quesada and Gheco 2015) and creating a common identity through the reuse of places, by reorganizing them and giving them new contents and meanings (Armstrong Bruzzone 2012; Recalde and Colqui 2022; Wrigglesworth 2006). In these interpretations, the complete or advanced overlapping of motifs tends to be associated with destroying behaviors and implies attitudes of forgetting, denying or annulling the past. On the other hand, minimal superimpositions suggest continuity with the past and attitudes of respect towards the previous populations who made rock art. Although these propositions are heuristically relevant, informed studies show that they do not always work so straightforwardly. One example corresponds to the initiation ceremonies among the Cheŵa women from the woodlands of South-Central Africa observed by Zubieta (2022). On these occasions, they made animal-shaped low reliefs with unfired clay and painted them with white, red and black dots. These figurines resembled nearby pictographs and worked as mnemonic devices associated with specific teachings and songs. All the objects were destroyed at the end of the ceremony, and the residues were thrown in a secret place. Destroying those ephemeral material forms required making them again and, thus, provided women an opportunity to remember teachings through their repeated production (Zubieta 2022). Another case concerns the Wandjina rock art from the Kimberley Plateau in Australia. Wandjinas are anthropomorphic figures, generally painted on white backgrounds, that resemble heads with eyes and a nose but without a mouth. This rock art corresponds to a painting tradition that goes back to at least 4000 cal years BP (O’Connor et al. 2022). The Worrorra, Wunambal and Ngarinyin clans from this region describe themselves as the Wandjina Wunggurr people and consider the Wandjinas as their ancestors. As creators of people’s cultural laws and landscape, these beings guaranteed life renewal to their descendants if they were properly attended to by looking after the land. Taking care of the pictographs involves knowing the songs, dances and stories connected with the creation of the landscape and repainting them. What is relevant in this example is that repainting may involve refreshing the whole image or a portion with new paint or modifying the image by disregarding aspects of the original. Evoking the creation narratives further involves scratching and pecking the Wandjinas, as well as drawing over them with charcoal (O’Connor et al. 2022). Rather than disrespect or denial of the past, these attitudes are associated with maintaining and passing on their traditional knowledge.

2. Regional Background

The human occupation of Southern Patagonia began in the transition from the Pleistocene to the Holocene, when the last glacial advances and retreats took place (

Table 1. Chronology of the main sites with rock art from the Deseado Massif and Río Pinturas basin.

14C Years BP	Temporal Block	Río Pinturas	Deseado Massif
c. 500	Post-European times Recent	Cueva de las Manos	Viuda Quenzana Los Navarros
c. 1000	Final Late Holocene		Los Toldos: Cueva 1
c. 3500	Initial Late Holocene	Cueva de las Manos Alero Rosamel	La Gruta Viuda Quenzana Cueva Moreno Cueva de la Hacienda Cueva Maripe
c. 7500	Middle Holocene	Cueva de las Manos Arroyo Feo 1 Charcamata 2	Los Toldos: Cueva 3 and Cueva 13 Piedra Museo Cueva Maripe La Gruta La Martita: Cueva 4 Cerro Tres Tetas La María: La Mesada and La Ventana
c. 9500	Early Holocene	Cueva de las Manos Arroyo Feo 1 Alero Cárdenas	Los Toldos: Cueva 3 Cueva Maripe El Verano: Cueva 1 La Martita: Cueva 4 La María: La Mesada
c. 11.500	Final Pleistocene		Los Toldos: Cueva 3 Piedra Museo Cerro Tres Tetas La María: Casa del Minero and Cueva Túnel La Gruta

Sources: Aguerre (1987); Aguerre and Gradin (2003); Alonso et al. (1984–1985); Aschero et al. (2019); Blanco (2015); Brook et al. (2018); Cardich et al. (1973); Durán (1986–1987); Franco et al. (2010); Gradin and Aguerre (1994); Gradin et al. (1976, 1979); Miotti (1998); Miotti and Salemme (2005); Miotti et al. (1999, 2005, 2014, 2022, 2024); Paunero (2000, 2003a, 2003b); Paunero et al. (2005, 2017, 2020); Rabassa et al. (2022). The dotted line represents the Pleistocene/Holocene transition. Cueva 2 is not included in Table 1 because it lacks radiocarbon dates. According to Menghin’s (1952) geochronological estimations, the cultural layers he defined as Toldense and Casapedrense presumably correspond to the Early Holocene and Middle Holocene, as it was corroborated in Cueva 3 (Cardich and Paunero 1991–1992).

). In this period of environmental instability, the colonizing populations were small, highly mobile and dispersed in the region. One of the most important concentrations of early archaeological sites (base camps, kill and butchering loci and workshops) is in the Deseado Massif (Figure 1). This plateau was attractive because it offered good-quality lithic raw materials, abundant natural shelters, fresh water, fuel and plenty of game. These conditions stimulated the permanence of people for several generations and the establishment of regional continuity (Miotti 1998; Miotti and Salemme 2004; Politis and Borrero 2024). The zooarchaeological record of Cueva 3 (Los Toldos), Alero El Puesto (Piedra Museo), Casa del Minero and Cueva Túnel (La María) points towards the consumption of Pleistocene mammals, such as Hippidion saldiasi (native horse), Mylodon sp. and ancient camelids (Hemiauchenia paradoxa and Lama gracilis) (Miotti and Salemme 2005; Paunero 2003a; Paunero et al. 2017, 2020). These animals were complementary resources to guanacos (Lama guanicoe), which were the most popular prey among the early groups. Following deglaciation and temperature increases, a series of environmental changes had an impact on the early populations. The most important were the rise in sea level and the reforestation of the Andean area, both of which modified the available land. The extinction of the Pleistocene mammals affected subsistence strategies, which focused more heavily on guanaco (Martínez et al. 2016; Miotti et al. 2018). This shift was associated with the replacement of the fishtail projectile points by triangular-shaped points (Hermo et al. 2022; Politis and Borrero 2024). The increase in the number of radiocarbon dates and occupations from the Early Holocene indicates a higher population density (Table 1). Places were used more redundantly, and the social landscape changed towards more defined territorial systems (Borrero 1999; Miotti and Salemme 2004; Miotti et al. 2022; Mosquera 2018).

The Mid Holocene environment is characterized by a temperature increase and less continental conditions. The known utilization of the coasts began in this period. In mainland Patagonia, there is evidence of a relatively more continuous regional occupation of most ecosystems (Politis and Borrero 2024). The archaeological signal of the Deseado Massif increases through a larger number of sites, with evidence of more intense and continuous occupations (Table 1). Guanacos continued to be the main prey and were intensively captured with different weapons, such as triangular projectile points and bolas. The period is also distinguished by the widespread production of blades (Hermo and Magnin 2012; Miotti et al. 2014). During the Mid and Late Holocene, the larger presence of foreign elements such as obsidian, cordilleran reeds and beads on sea snails, limpets and bivalves suggest that hunter gatherers established more solid territorial relationships and wider social interaction networks (Miotti 2006; Miotti et al. 2024). The Late Holocene tends towards the configuration of the present Patagonian landscape through an aridification process and volcanic episodes. Sites become larger, sometimes indicating multiple activities. The increment in the number of sites and their wider distribution in the landscape suggest a demographic increase. Residential mobility is lower, and logistical mobility is higher, as implied by the larger frequency of small sites (Miotti 1998; Borrero 1994–1995). Human populations manage to establish temporary hunting stations on the high plateaus near the Andean foothills, where stone hunting blinds and petroglyphs are abundant (Belardi and Goñi 2006; Goñi 2010; Cassiodoro et al. 2014). The main technological innovations registered for this period are pottery and the bow and arrow (Politis and Borrero 2024). The engraving technique is introduced in rock art production and becomes especially intensive in open-air sites (Fiore 2006; Fiore and Acevedo 2018; Re 2010). Social networks and communication between different territorial systems are amplified (Goñi et al. 2019; Miotti 2012; Re et al. 2024).

Throughout this process that lasted millennia, many caves and rock shelters from the Deseado Massif and Río Pinturas basin were redundantly occupied. This is evidenced by their long stratigraphic sequences and the large palimpsests of pictographs concentrated in their interior. Among the most important are Cueva de las Manos, Arroyo Feo 1 (Aschero and Schneier 2023; Aschero et al. 2019; Gradin et al. 1976, 1979), Los Toldos (Cardich et al. 1973; Carden and Miotti 2020; Carden et al. 2018; Miotti et al. 2024), Cueva Maripe (Carden 2008; Miotti et al. 2014), Piedra Museo (Carden 2008; Miotti et al. 2022), La María (González Dubox 2024; Paunero 2003a; Paunero et al. 2005), Cerro Tres Tetas (Paunero 2003b), El Ceibo (Cardich 1979), La Martita (Gradin and Aguerre 1983), El Verano (Durán 1983–1985) and La Gruta (Acevedo 2022; Fiore et al. 2017; Franco et al. 2010) (Figure 1 and Table 1).

3. Patagonian Rock Art in Post-Contact Times

In the 18th century, the name “Tehuelche” designated various ethnic groups who inhabited continental Patagonia (Nacuzzi 1998). The Aónikenk, or Southern Tehuelche, and the Gununa’kena, or Northern Tehuelche, were the last representatives of a long tradition of guanaco hunters who used the bola stone and the bow and arrow in their hunting practices (Casamiquela 1969; Escalada 1949; Politis and Borrero 2024). There are a few references to rock art in the Tehuelche’s mythology (Bórmida and Siffredi 1969–1970; Gradin 1971; Wilbert and Simoneau 1984). However, there are no direct or indirect accounts of this people making images on rocks (Schobinger and Gradin 1985). Musters ([1871] 2007, p. 247) provided information on the healing ritual for a sick child among the Southern Tehuelche (Aónikenk), in which a white mare was sacrificed and stamped with hands covered in red ocher. This practice was interpreted as the persistence of the ancient habit of producing hand stencils and handprints in the caves and rock shelters of South Patagonia (Menghin 1957). Francisco Moreno was the first explorer who reported the existence of rock art in Patagonia and described some of the depictions (Fiore and Hernández Llosas 2007). His judgment of the pictographs of Lago Argentino (Santa Cruz province) reveals a racist evolutionist perspective, according to which the Tehuelche were in a stage of involution from former cultures. These inscriptions were “…the testimony of the existence of more perfect and moral men than the Tehuelche Indians, who had no other idea of drawing than the shapeless lines and dots they made on the reverse of their cloaks (quillangos)” (Moreno [1877] 2007, p. 93, our translation). In several instances during his excursions to North Patagonia, the Northern Tehuelche (Gununa’kena) told Moreno about a bad spirit called Ellengasem (popularly known as Gualicho). According to different versions, this being was a giant with a human face and an armadillo shell, who inhabited caves and painted figures in their interior (Casamiquela 1960). Since these places were feared and avoided, their accounts contributed to Moreno’s argument about the relationship of rock art to a remote and extinct race who had preceded the contemporary Indigenous people. This promoted a disconnection between rock art and ethnographic people that persisted in academic thought. In reference to this mythological creature, who was the creator of rock art, Casamiquela (1960, p. 10) stated that “… fantastic explanations do not arise through a generation. On the contrary, they are the product of a hiatus in historic continuity…” (our translation).

Within the culture–history framework, the latest rock art styles were interpreted as the artistic expressions of the Tehuelche’s ancestors (Menghin 1957; Gradin 1976). However, most of these researchers agreed about a discontinuity between the prehistoric and ethnographic information, because there were no accounts of Indigenous people producing rock art in post-contact times. Despite paradigmatic change, this idea was maintained from processual archaeology, since the archaeological record showed that rock art was generally located in the best areas of shelter, water and animal and vegetal resources, and it shared the same space with other vestiges of people’s daily life. Such a pattern provided a completely different picture from Moreno’s reports, according to which rock art sites were feared and avoided by the Tehuelche (Aschero 1996; Aschero et al. 2005). Despite the disconnection perspective, Tehuelche myths and their cosmovision have been recently considered to interpret the meaning of certain motifs and panels from Cueva de las Manos. Aschero and Schneier (2024) argue in favor of using a direct historical analogy, because the Southern Tehuelche, as descendants of the original populations who inhabited the same region, shared enough information about their beliefs.

The lack of ethnographic information about the production of rock art can be understood in the context of the progressive social fragmentation of the Tehuelche’s worlds after the European conquest and re-colonization of what was claimed to be a poorly inhabited desert (Briones and Delrio 2007). This rupture may have led to the disappearance of the practice of making rock art, but it is also true that it is the product of the prejudices and biases of the voyagers and ethnographers who contacted them, as well as of the researchers who used their chronicles. The information about Ellengasem can be interpreted in a different way in the light of indigenous ontologies (Alberti 2016). In other ethnographic contexts where rock art has been made until recent times, like the Australian, when people were interviewed about their motivations for painting rocks, they did not consider themselves as the authors. The pictographs were not representations of something; they were in themselves the manifestation of spirits or ancestors, and the people’s role was repainting them and interacting with them through songs, stories, performances and offerings (see Bradley et al. 2021; Domingo Sanz 2021; O’Connor et al. 2022; Vinnicombe 2010 for examples from the Northern Territory, the Kimberley Plateau and Western Australia). Given that the author of rock art was also an other than human being in Moreno’s report, whether the Tehuelche produced rock art or not is irrelevant from this perspective, since it is evident that they interacted with it and incorporated it into their social memory. Some myths documented by Bórmida and Siffredi (1969–1970) from several Southern Tehuelche (Aónikenk) elders are illustrative on this point. In different versions referring to Elal, who was the creator of the Tehuelche and their culture, they mention an event in which he asked permission of the Sun and the Moon to marry their daughter. As a response, he had to undertake very dangerous tasks, all of which he accomplished. They tried to trick him, disguising a servant as their daughter, but Elal realized that she was not his claimed bride. As a reprisal, he did not marry their daughter and sent them to the sky. According to the different versions of the myth, when the Sun and the Moon took off to the sky, their hut turned into stone, and their fingerprints, handprints, footprints and genitals remain engraved on the dwelling’s surface. Furthermore, the elder informants mentioned specific places where these traces could still be observed. In certain versions of Elal’s myth, the hero departed to the sky riding a swan. His flight was not straight but “as a wheel”, according to Ana Montenegro de Yebes’ description (Bórmida and Siffredi 1969–1970, p. 215), which referred to a spiral trajectory. These fragments of Tehuelche’s mythology recall certain rock images (especially petroglyphs) and motif associations from the Deseado Massif and beyond in Santa Cruz province (Acevedo et al. 2012–2014; Carden 2008; Gradin 1976, 1996; Re 2010, among others). This does not suggest a continuity in meaning through thousands of years, since material culture is polysemic, and symbols are not inert (Hodder 1982). Instead, it implies that rock art had a relevant role in the Tehuelche’s social memory, even though they were not its makers.

4. Los Toldos Locality

Los Toldos is situated in the volcanic landscape of the Deseado Massif on the extra-Andean Patagonian steppe (Figure 1). Las Cuevas canyon is a 1.5 km long ignimbrite formation, along which a temporary stream runs towards the Deseado River. This sheltered microenvironment offered good conditions for human settlement in an arid region where water was a critical resource. At present, fresh water comes from the temporary stream and from springs between 2 and 5 km from the central caves. Among the 16 caves and rock shelters identified in Las Cuevas canyon, 11 include pictographs (Figure 2).

The locality has been known in the academic field since the first half of the 20th century (De Aparicio 1935). In the 1950s, it acquired a relevant position as a testimony to the ancient cultures who inhabited Patagonia (Menghin 1952). The first radiometric ages were obtained after the excavations conducted in the 1970s (Cardich et al. 1973; Cardich 1984–1985). These results contributed to the discussion of the early peopling of the Patagonian region, since Cueva 3 yielded cultural evidence from the late Pleistocene. Even though there is strong criticism around these early dates, and Los Toldos needs a redating program (Borrero and Franco 1997; Miotti 2003; Mosquera 2018; Politis and Borrero 2024; Prates et al. 2013, among others), it continues to apply as a relevant locality for discussing the peopling process of Patagonia (Carden and Miotti 2020).

Since the initial stylistic sequence proposed by Menghin (1957), rock art from Los Toldos has been understood as a phenomenon that lasted several millennia. From the analysis of the superimpositions in Cueva 2 and Cueva 3 through a Harris matrix, Carden and Miotti (2020) defined a rock art sequence of eight episodes. Considering the local archaeological context and the regional information from other archaeological localities with redundant occupations and rock art, these episodes were broadly related to three temporal blocks (

Table 2. Rock art sequence of Cueva 2 and Cueva 3, Los Toldos.

Temporal Block	Cueva 2	Cueva 3
Final Pleistocene	Episode 1	Episode 1
Early Holocene	Episode 2	Episode 2
Middle Holocene	Episode 3	Episode 3
	Episode 4	Episode 4
	Episode 5	Episode 5
Late Holocene	Episode 6
	Episode 7
	Episode 8

). The main tendencies of this sequence, which needs to be tested against a direct dating program of pictographs, are the following:

• The Pleistocene/Holocene transition is characterized by two painting episodes. The first one includes red and brownish-yellow motifs, which are mainly hand stencils and, to a lesser degree, dots and spirals. The second episode (Early Holocene) consists of black and black/red hand stencils, followed in lower amounts by black dots, lines and dashes. In Cueva 3, two episode 2 black hand stencils are eroded in their inferior parts by a series of natural parallel lines. These marks could correspond to a flood, according to Menghin (1952, 1957), or to the deposition of sediments, according to Cardich et al. (1973). Although their interpretations are dissimilar, both researchers place the natural event in the hiatus between the Toldense and Casapedrense stratigraphic layers. Since the episode 2 black stencils are previous to the natural event, they could correspond to the Toldense occupation, dated to c. 8750 years BP from a hearth in the top of layer 8. The low motif frequency in this temporal block is the result of the sporadic character of the occupations. Moreover, it is the consequence of the low visibility of the initial rock art episode due to its poor preservation and its position in the bottom of the painting sequence.
• The Middle Holocene is distinguished by the alternation of red and white painting episodes (3, 4 and 5). Since they are similar in motif composition, it is plausible that they were closely linked. Although the temporal distance between them is unknown, they represent the peak of rock art production and image variability. It is relevant to consider that the most intensive human occupations registered in Los Toldos through the excavations of Cueva 2 (layer 4), Cueva 3 (layers 6 and 7) and Cueva 13 (layers 6, 7, 8 and 9) were identified as Casapedrense. This cultural level was dated between c. 7260 and c. 4800 years BP in Cueva 3 (Cardich et al. 1973; Cardich and Paunero 1991–1992; Miotti 1998). In this period, although hand stencils continue to be predominant, their percentage decreases with respect to other motifs, such as geometric signs and footprints.
• The Late Holocene painting activities are only represented in Cueva 2 (episodes 6, 7 and 8). Although their frequency declines with respect to the previous period, they exhibit more diversity in subject matter and colors (yellow, brownish yellow, red, black and white). Hand stencils are no longer predominant among other images, such as geometric figures, handprints and bird footprints. The drop in motif frequency can be understood if it is considered that no Late Holocene human occupations were registered in the excavations of Cueva 2. Moreover, the top cultural layers of Cueva 3, which, according to their material culture, correspond to this period, are less intensive than the Middle Holocene ones (Casapedrense).

During the Pleistocene/Holocene transition and Early Holocene, hand stencils were the most abundant motifs in Cueva 2 and Cueva 3. In a social context in which the demography was low and populations were dispersed, Los Toldos was recurrently used by hunter gatherers within a programmed mobility circuit (Miotti 1998; Miotti and Salemme 2004). Hand stencils were relevant mediators of social interaction among those who were not in direct contact, because they were indexes of people and their movements across the landscape. Their position in the most visible panels enhanced communication (Carden and Miotti 2020). In their repeated visits to the site, people remembered placing their own hands or inferred who had been there through the hands’ size, shape and, probably, through their color. The superimpositions registered within the initial painting episodes of Cueva 2, which mainly involved hand stencils, were not a consequence of a lack of space in the cave. As “distributed persons” (Gell 1998), these motifs had a power of attraction (probably genealogical) that motivated the location of future hands1. The production of hand stencils continued through many generations, and ancestry became powerful. Thus, hand stencils became important social actors that participated in an intergenerational process of communication through which social memory was built (Carden and Miotti 2024).

From the Middle Holocene, hand stencils started to decrease with respect to other motifs, such as human and animal footprints and geometric figures. The codes for understanding the latter images (which were more than indexes) and their combinations with other motifs could have been taught in contexts of social fusion in which pictographs were observed collectively, as their high horizontal position on the ceilings allows. Other motifs, such as long lines extending through different panels or even through a whole chamber, had to be walked around to visualize them completely. Placing this rock art on high walls and ceilings required artificial means of elevation and group collaboration. These location patterns suggest that rock art’s production, subsequent observation and interpretation were shared activities and a means of strengthening social bonds in a period when the locality was intensively used (Carden and Miotti 2020). The absence of a Late Holocene human occupation in Cueva 2 and its lower signal in Cueva 3 may explain why rock art production became more infrequent in this period. However, this does not mean that Los Toldos was abandoned, since recent excavations in Cueva 1, 200 m east from Cueva 2, recovered a Late Holocene human occupation of c. 1200 years BP. This cultural layer contained lithic remains, animal bones, vegetal macro vestiges and pottery (Miotti et al. 2024). It is relevant that the pictographs registered in small niches outside Cueva 1 resemble those from the latest episodes of Cueva 2 (6, 7 and 8).

Measurements applied to the hand stencils of Cueva 2 revealed that they belong to adults and subadults throughout the Early, Middle and Late Holocene. Although adult hands are predominant, it is possible that the subadult sample may be underrepresented by a taphonomic bias. The predominant position of small hands in the lowest sections of the panels makes them more vulnerable to different agents such as cattle rubbing, sedimentation and water percolation (Carden and Miotti 2020). Unfortunately, the employed methodology did not allow sex to be discriminated, except in the largest hand stencils, which most likely belong to adult men (Carden and Blanco 2016; Gradin 1981–1982). The dotted line represents the Pleistocene/Holocene transition.

5. Materials and Methods

Among the 887 motifs registered in Cueva 2, 413 (47%) could be related to specific painting episodes (Carden and Miotti 2020, Table 1). The analysis conducted here relies upon these episodes and considers 191 cases of superimpositions to evaluate how motifs from each episode overlay previous images. Given that episode 1 pictographs are on the base of the sequence, the analysis considers episodes 2 to 8. Based on Re (2016), four kinds of superimpositions are defined:

• Minimal: when the superimposition covers less than 10% of the underlying motif;
• Partial: when the superimposition covers more than 10% and up to 50% of the underlying motif;
• Advanced: when the superimposition covers more than 50% of the underlying motif, which remains partially visible;
• Complete: when the superimposition covers the complete area of the underlying motif (or most of it).

To obtain clear tendencies, minimal and partial superimpositions were grouped into Class 1 superimpositions, in which less than 50% of the underlying motifs’ surface is covered. On the other hand, advanced and complete superimpositions were grouped into Class 2 superimpositions, in which the underlying motifs are covered for more than half of their surface. Class 1 superimpositions imply an intention to leave the underlying motifs visible, while Class 2 superimpositions lack this concern. Furthermore, we considered other temporal relationships, which may or may not involve superimposing images (Aschero 1988; Re 2016):

• Maintenance: when a motif is repainted without changing its original morphology;
• Recycling: when a motif is repainted or new elements are added to it (or close to it), and a new image or composition is created;
• Replication: when a motif is copied in a later episode, although its color may be different.

Recycling is easier to visualize than maintenance, because it implies modifying the original image. It was possible to recognize maintained pictographs when they exhibited brighter colors or thicker layers of paint compared to similar motifs located on the same panels.

6. Results

6.1. Kinds of Superimpositions in Cueva 2

This stage of the analysis evaluates how motifs from each painting episode overlay previous images. Except for episode 8, which only includes two cases, the remaining episodes show a predominance of Class 1 superimpositions, which are most abundant in episodes 4, 5 and 7 (

Table 3. Kinds of superimpositions along the painting sequence of Cueva 2.

Temporal Block	Episode	Class 1	Class 2	Total
Early Holocene	2	11 (65%)	6 (35%)	17 (100%)
Middle Holocene	3	16 (62%)	10 (38%)	26 (100%)
	4	45 (82%)	10 (18%)	55 (100%)
	5	26 (76%)	8 (24%)	34 (100%)
Late Holocene	6	24 (65%)	13 (35%)	37 (100%)
	7	16 (80%)	4 (20%)	20 (100%)
	8	1 (50%)	1 (50%)	2 (100%)
Total cases		139 (73%)	52 (27%)	191 (100%)

). When the episodes are grouped into temporal blocks, an increase in Class 1 superimpositions is observed towards the Middle Holocene, when they peak, to decline again in the Late Holocene (Figure 3). However, Class 1 superimpositions remain numerically abundant with respect to Class 2 superimpositions along the whole sequence. This tendency implies that the previous images were relevant to ongoing generations of painters.

The relative abundance of Class 1 superimpositions (minimal and partial) suggests that the main concern in Cueva 2 was integrating the new images with the older ones. Despite this tendency, Figure 4, Figure 5 and Figure 6 show a much more complex situation in which there were no strict rules for superimposing. The presence of Class 2 superimpositions (advanced and complete) in the first painting episodes was not a consequence of a lack of space available for painting inside the cave. These kinds of superimpositions also occurred in later episodes, when the dark niches at the back of the cave began to be painted (E.6 and E.7 in Figure 6B). However, the most recurrent behavior in Cueva 2 consisted of placing pictographs over the already densely painted panels.

6.2. Other Relationships with Previous Rock Art in Cueva 2

The rock art sequence of Cueva 2 did not only involve producing new images but also repainting the already existing ones, whether to maintain their original morphology and color or to modify them through recycling. An example of recycling behavior is a polychromic figure created through the minimal superimposition of black dots (E.7) over a purplish red dotted circle (E.3), which was in turn completely superimposed over red clustered dots (E.1) (Figure 7A,B). Painting (or repainting) in red the interior of white hand stencils (E.4) was a common way of maintaining motifs or recycling them (e.g., Figure 4C and Figure 7C, respectively). Another case is represented by a black hand stencil (E.2) that was repainted by placing a hand of similar size over it, spitting white paint to make a new contour and painting the palm red. Through a Class 2 superimposition, the black hand stencil became obliterated and transformed into an E.4 white hand stencil (Figure 8A,B). Other recycling events involved minimal superimpositions, such as a black hand stencil (E.2) that was incorporated into a composition of a red circle and two red hand stencils (E.3), over which a white hand stencil (E.4) was added through a partial superimposition (Figure 8C).

Another form of interacting with the past in Cueva 2 involved reproducing the same classes of motifs over millennia. Hand stencils, dots and lines are the most common examples throughout the painting sequence. Stenciled guanaco footprints and stenciled human feet start to be produced in the Mid Holocene, and handprints and bird footprints are incorporated in the Late Holocene. These motifs continue the modality of representing animals and humans through their prints, as hand stencils did from the beginning of the site’s occupation. Not only recurrent motifs but also more specific images were copied through time, such as three purplish-red thin spirals (E.1) that were replicated in pink several millennia later (E.5) (Figure 9).

7. Discussion

The predominance of Class 1 superimpositions (minimal and partial) in Cueva 2 shows a concern for leaving previous images visible and interacting with them. This implies that they were relevant to the social memory of the people who successively inhabited the place. However, these results should not obscure the fact that Class 2 superimpositions also occurred along the sequence (Figure 3). This is especially outstanding in the first painting episodes, when there was plenty of space available for painting inside the cave. Rather than breaking with the past through annulling the previous images, advanced and complete superimpositions may be suggesting intentions to be connected to the preceding pictographs, even though some of them became obliterated. Although these behaviors are difficult to understand from our Western logic, they could have been a means of reaffirming social ties with the preceding populations. Class 1 superimpositions are especially manifested in the Middle Holocene (Figure 3). If it is considered that the peak of rock art production occurs in this period, which coincides with the most intensive occupations of Los Toldos, the predominance of minimal and partial superimpositions shows a special concern for interacting with the past through leaving the images visible. As mentioned above, the position of these pictographs on elevated ceilings and walls required artificial means of elevation and group collaboration to produce them. Their high position also allowed them to be observed collectively (Carden and Miotti 2020). It is relevant to consider that, regionally, wider social interaction networks and more solid territorial relationships were registered among the Middle Holocene human populations (Borrero 1994–1995; Miotti and Salemme 2004). In this context, the interaction with the past, as manifested through these superimpositions, could have been especially important for building an ancestral memory and strengthening bonds in contexts of social fusion. The less intensive rock art production that occurred during the Late Holocene, as well as the continuity in the predominance of Class 1 superimpositions, can be explained by the large palimpsest of images that was already present and still had transcendental meaning in the social memory of the latest generations who inhabited Cueva 2.

A common way of recalling the past in the studied site was replicating the same classes of motifs throughout time. This behavior is not particular to Cueva 2 but characterizes much of the rock art sequences in the region and beyond. The long-term persistence of images was detected in Viuda Quenzana, 150 km south of Los Toldos in the Deseado Massif, where Acevedo (2022) performed a detailed analysis of superimpositions in 11 sites. According to the chronology obtained from the excavation of two sites (Brook et al. 2015) and the direct dating of two pictographs (Brook et al. 2018), the human occupation of this archaeological locality occurred between c. 4700 and c. 500 years BP. However, some painted guanacos suggested an older age through their style (7300–5300 years BP). The analysis revealed a tendency towards the long-term production of the same classes of motifs—i.e., hand stencils, footprints, guanacos and simple and complex geometric figures—by the painting technique (Acevedo 2022, Figure 2a). Since information degrades with time elapsed, Acevedo interpreted the diachronic motif repetition as an effective strategy for storing and conveying information through generations in a context of environmental uncertainty. In the Río Pinturas area, Cueva de las Manos has a rock art sequence that began c. 9400 years BP and continued until c. 500 years BP (Gradin et al. 1976). The first 2600 years of this sequence are characterized by the depiction of guanaco hunting scenes associated with hand stencils painted in the same colors, such as ocher, red, black, yellow and white. According to Aschero and Schneier (2023), these pictographs had an educational function for those who were initiated in hunting practices and manifested a deep concern with the origin, reproduction and hunting of guanacos. Although far from our study area, it is relevant to mention the pictographs from Cueva Huenul, 1200 km northwest of Los Toldos in Northern Patagonia (Neuquén province). The radiocarbon dates obtained from three comb-shaped black motifs showed that this image was replicated from 6830 ± 50 to 4730 ± 110 years BP, calibrated to 7565–7728 and 5629–5643 years BP, respectively. A fourth comb-like black motif yielded an age of 3010 ± 160 (calibrated to 2761–3477 years BP), although it was rejected by Romero Villanueva et al. (2024, Table 1) because the sample was too small. The persistence of the image throughout time was interpreted by the authors as a sustained intergenerational transmission of Indigenous knowledge during the Mid Holocene. Thus, the maintenance of collective memory was a resilient response to ecological stress in an arid environment.

The maintenance and recycling of images, as well as their integration into new compositions, are other common behaviors in Cueva 2 that are repeated in other sites from Southern Patagonia. In Cueva de las Manos, new hunting scenes were made with different colors to distinguish them from the older ones. Instead of obliterating the underlying images, people integrated their new productions into those previous compositions. As they interacted with preexisting rock art, maintaining the motifs in sight and sometimes repainting them, Cueva de las Manos became a place of reference because of what their ancestors had painted in the past. In this way, rock art activated a social memory of their remote past, and Cueva de las Manos became a “collective memory archive” (Aschero and Schneier 2023, p. 108). From the analysis of the superimposed petroglyphs from the Strobel Plateau (Central-West Santa Cruz province), Re (2016) found that in most superimpositions, the top motifs only covered a small portion of the underlying images. This pattern implied an intention to keep the previous images visible and to continue interacting with them. Furthermore, it showed an attitude of respect towards previous rock art, which remained relevant for the ongoing generations from the Middle Holocene to the post-contact period. Papú (2023) arrived at similar conclusions through her study of the superimposed pictographs in the Cerro de los Indios locality, Northwestern Santa Cruz. Her results suggested that from c. 3800 to c. 2500 years BP, pictographs were produced in high panels and related to the underlying motifs through recycling elements of these previous images to incorporate them into new designs. These kinds of attitudes towards older images are not restricted to Patagonia. The Cerro Colorado locality, situated in the northern hills of Córdoba province (Central Argentina), displays a large concentration of pictographs, distributed in 60 sites. These images were produced throughout periodic visits to the place during the Late Prehispanic period, from c. 1200 to c. 450 years BP. Through their study of superimpositions, Recalde and Colqui (2022) interpreted the rock art panels as “open works”. Even though the addition of new motifs through minimal superimpositions did not affect the previous ones, repetition did not prevent variation or the creation of new meanings.

The examples mentioned above show that people chose to preserve rock art and tended to keep it visible, and this phenomenon kept an ancestral memory alive. However, the preserved images were subject to interpretation, which could change over time, and their accumulation allowed the manipulation of the information that they originally transmitted. In this sense, memories were created, modified and reinforced (Whitley 2022). Our analysis of superimpositions in Cueva 2 suggests that memory was built in this way, particularly through the integration of images from different times into narratives. This argument is supported by a comparative analysis with the rock art repertoire of other sites that formed part of the mobility circuits of the hunter gatherer societies who inhabited the Deseado Massif since the Pleistocene/Holocene transition. A relevant example is constituted by the petroglyphs of the Piedra Museo locality, 80 km southeast of Los Toldos (Figure 1). Very similar motif associations, also known as “themes” (Aschero 1997), were registered between one of the engraved boulders of the Cueva Grande site (Piedra Museo) and a painted panel from Cueva 2. Through the analysis of superimpositions, it was established that the associated motifs in Cueva 2 (panel 40) were not produced synchronically. Some of them correspond to painting episodes 1 and 2, linked at least to the Early Holocene, and others belong to episodes 3 and 4, related to the Mid Holocene (Carden and Miotti 2020) (Figure 10A,B). On the contrary, the similar motif associations found in Cueva Grande (boulder 2) were presumably made synchronically, according to the lack of superimpositions and their similar techniques, groove depth and patina (Figure 10C,D). The estimated chronology of these petroglyphs corresponds to the late Mid Holocene or initial Late Holocene (Carden 2008, 2022). The presence of similar themes in the region suggests the circulation of shared narratives through the landscape (Aschero 1997; Carden 2008). However, this does not necessarily imply synchronicity in the production of the painted and engraved images. In the present case, it shows that in the Mid/Late Holocene, motifs painted in different episodes in Cueva 2 were integrated with narrative purposes. These stories were replicated through the production of similar motif associations in the context of people’s movements and/or social interaction networks.

The ethnographic and ethnohistoric record from the post-contact period in Patagonia lacks information about the production of rock art. However, it is relevant that one of the most conspicuous mythical beings reported to Francisco Moreno by the Northern Tehuelche (Gununa’kena) was known to be the maker of rock art (Casamiquela 1960). This human/armadillo-like monster, named Ellengasem, may be referring to Glyptodon sp., whose fossils were known and collected by these groups (Bonomo 2006; Casamiquela 1988). The presence of animals resembling Pleistocene mammals in Holocene rock art contexts from the Deseado Massif and Río Pinturas suggests that these animals survived in the social memory of the hunter gatherers who inhabited the Patagonian plateaus. Examples of motifs that resemble extinct fauna are a large spotted feline (Panthera onca mesembrina) in El Ceibo, unshod horse footprints (Hippidion saldiasi) in Piedra Museo and a zooanthropomorphic figure similar to an upright armadillo with a spiky long tail (possibly Glyptodon sp.) in Cueva de las Manos (Carden 2009; Cardich 1979; Miotti and Carden 2024). According to the myths recorded by Bórmida and Siffredi (1969–1970), rock art from Southern Patagonia continued to be a relevant mnemonic device in the 20th century. Even though the Southern Tehuelche (Aónikenk) were never documented making rock art, the interviewed elders remembered stories about the creation of their culture that were tangible in the landscape they inhabited. Certain elements from the myth about Elal (their hero) and his confrontation with the Sun and the Moon recall specific rock images. One example is the presence of hands, feet, dots (fingerprints?) and a spiraled figure (Elal’s flight?) in Piedra Museo (Figure 10C,D). These similarities do not imply that the meaning remained intact since the images were produced. What seems to be continuous in the Deseado Massif, at least as regards the sites analyzed here, is the interaction with the past, to build a social memory in an inherited landscape whose origins were always recalled.

8. Conclusions

This work conceptualizes rock art as a means of commemoration that has passed through different social systems through its perdurability (Armstrong Bruzzone 2012). Many localities from the Deseado Massif have been occupied in the long term, and their cumulative rock art palimpsests (sensu Bailey 2007) have been disentangled into long sequences (e.g., Acevedo 2022; Aschero and Schneier 2023; Carden and Miotti 2020; Gradin et al. 1976). The detailed analysis of superimpositions to identify human attitudes towards previous rock art is a crucial avenue for detecting connections between different temporalities. Previous studies in Los Toldos characterized the long-term human occupations as different cultures (Menghin 1952; Cardich et al. 1973) or as populations who changed their subsistence and settlement strategies (Miotti 1998). The hiatuses found in the stratigraphic layers of Cueva 2, Cueva 3 and Cueva 13 show that the occupation of Los Toldos was not a continuous process (Menghin 1952; Cardich et al. 1973; Miotti 1998; Miotti et al. 2024). However, the locality overall applies as a persistent place (sensu Schlanger 1992) that remained attractive during millennia for the human populations who redundantly inhabited its different caves and rock shelters. After recurrent visits to Los Toldos, people remembered specific events, ancestors and related places, through which they built their genealogical history. Simultaneously, they recalled a remote past through their mythological history (Gosden and Lock 1998). Both histories were intimately intertwined and shaped their lives and movements. This work places emphasis on the users/consumers of rock art rather than on its makers (Motta 2019; Norder 2012). Such an approach narrows the gap between the archaeological and ethnographic information about Patagonian rock art. However, this was not achieved through a direct historical analogy to interpret meaning, assuming a continuity in belief among people who were no longer related to this materiality. On the contrary, this contribution claims that the Tehuelche people continued to be engaged with rock art as a manifestation of other than human beings with whom they interacted and built their social memory. Future work in this direction remains to be carried out in other caves from Los Toldos and other sites from the study area. This interpretive challenge will enhance our comprehension of the links between different places and times, from which the past was constantly renewed into the present and social memory was constructed in Patagonia.