The Camarones Basin

The Camarones River basin in northernmost Chile spans 4637km², extending 127km from east to west and 52km from north to south (Figure 1). This watershed encompasses Chiza, part of Nama and several smaller streams that feed into the main Camarones River. The Western Valleys, including the Camarones basin, feature two distinct ecoregions: the Atacama Desert to the west, one of the driest places on Earth, and the Dry Puna to the east, with peaks approaching 6000masl. In their longitude, the Western Valleys form a steep staircase-like landscape from sea level to the high mountain Puna—a climb dramatically interrupted by ravines that reach depths of 1000m.

Figure 1. (Top) The Camarones Basin (red polygon) in the context of the Western Valleys; bottom) archaeological sites identified within the Camarones Basin prior to this work. Contour lines are shown at 1000m intervals on both maps (figure by author, basemap: Esri World Imagery).

Despite the extreme aridity of the Atacama Desert, perennial and seasonal waters from the Puna provide essential hydration to the region. Seasonal monsoon rains cause natural fluctuations in the river’s flow, substantially impacting the distribution of flora and fauna in the Camarones basin and adjacent valleys, with amplified terrestrial biomass in the Puna (more than 3000masl) and rich marine resources by the Pacific Ocean (Figure 2).

Figure 2. Vegetation in the Camarones Basin during wet and dry seasons, based on above-ground vegetation analysis using the vegetational index MSAVI on 2018–2021 cumulative Sentinel-2 imagery (10m/pixel). Red indicates vegetation available in both the dry and wet seasons, while green indicates vegetation that is only available after the Andean rainy season (March–April) (figure by author).

Historical and cultural context

Archaeological evidence is not the only source of information for researchers aiming to interpret the emergence of increasingly complex societies in the Western Valleys after the Formative Period. Ethnohistory has played a crucial role in integrating chronicles and colonial documents with archaeological knowledge, particularly relevant for the eleventh century AD onwards. Yet the accounts of ethnic groups in colonial records are embedded within a context of exploitation under the Spanish Crown in the Americas. As such, the information they contain often reflects a logic driven by profit and extraction rather than by any neutral documentary intent, making it especially difficult to trace lower-status groups, such as the Uru, using colonial sources alone (Julien Reference Julien1987: 53). Much of the information recorded by Spanish colonists concerns ethnic groups of food producers that politically organised themselves in rank societies and chieftaincies (Nielsen 2006). Thus, the Late Intermediate and Late periods are nourished by information from highly hierarchical groups living under well-delineated socio-political regimes. As a result, the question arises as to whether the ethnic groups portrayed by the standard archaeological and ethnohistorical paradigms were the only ones encountered by Spanish settlers when they arrived in the Western Valleys. Fragmentary descriptions of people portrayed as maroon (cimarrones), poor, primitive and barbaric suggest that the ethnic tapestry was far more diverse than described by the archaeology of the late pre-Hispanic periods (Bouysse-Cassagne Reference Bouysse-Cassagne and Cook1975, Reference Bouysse-Cassagne1987; Julien Reference Julien1983; Bouysse-Cassagne & Harris Reference Bouysse-Cassagne, Harris, Bouysse-Cassagne, Harris and Platt1987; Wachtel Reference Wachtel2001). The most common name given to ethnic groups was Uru. The Uru, Uro or Huru have been defined in the colonial context as a specific tributary category (Julien Reference Julien1987; Bouysse-Cassagne Reference Bouysse-Cassagne1988: 76; Gil García Reference Gil García2013) that linguistically and economically generalises people who hunt, fish, who do not engage in farming and are particularly difficult to govern in the colonial system (e.g. de Matienzo Reference de Matienzo1967: 267). In the South-Central Andes, the Aymara, through their numerous chiefdoms, dominated the region and maintained close relations with the colonisers due to their territorial and productive power (Bouysse-Cassagne Reference Bouysse-Cassagne1987). Among the Pacaje, the dominant Aymara chiefdom in what is today the northernmost part of Chile, ethnohistorical accounts from the late nineteenth century indicate that hunters and fishers were held in low esteem, defined as “[a] nation of Indians despised by everybody, usually fishers, and of less understanding […] to one who walks dirty, ragged, rustic” (Bertonio Reference Bertonio1879: 380, author’s translation). This antipathy to populations understood today as foragers is upheld in the Andes with the Aymara concept chullpa puchu, which is used to refer to the Uru-Chipaya of the Coipasa salt flats and may be translated as ‘the scraps of the Chullpa’, a primitive society found before the time of the sun in the Aymara and Uru-Chipaya cosmogony (Wachtel Reference Wachtel2001: 15).

Despite the negative connotation of chullpa puchu, its meaning unfolds a path to trace colonial hunter-fisher-gatherer populations. In the current Aymara worldview, chullpa jaqui refers to ancestral people who play a role as souls that serve as mediators between the civilised society and the liminal world of wild forces (Bouysse-Cassagne & Harris Reference Bouysse-Cassagne, Harris, Bouysse-Cassagne, Harris and Platt1987: 40). Similarly, chullpa in the Aymara cosmogony forms part of the set of words linked to the Puruma or Purum pacha, a time before agriculture, before the sun with darkness only lessened by the moonlight, before the time of the sun and order (Bouysse-Cassagne & Harris Reference Bouysse-Cassagne, Harris, Bouysse-Cassagne, Harris and Platt1987: 23). One of the synonyms for Purum pacha in Aymara is chuquila or choquela. This multifaceted word, like chullpa, is associated with wild supernatural forces and cursed places, but was also used to refer to vicuña (Vicugna vicugna) hunters (Bouysse-Cassagne 1987: 89, Reference Bouysse-Cassagne1988: 36; Bouysse-Cassagne & Harris Reference Bouysse-Cassagne, Harris, Bouysse-Cassagne, Harris and Platt1987: 24; Wachtel Reference Wachtel2001: 23). Choquela has been linked with the natural force of lightning (Bouysse-Cassagne & Harris Reference Bouysse-Cassagne, Harris, Bouysse-Cassagne, Harris and Platt1987: 44), and in the Central Andes this word refers to a ritual dance that enacts the hunting of vicuñas using a trap with sticks and ropes (Gutierrez Reference Gutierrez1992), alluding to the funnel-shaped chacu traps.

Although choquela could be interpreted as a blanket term for hunters, research addressing the problem of foragers in the region during colonial times concedes that the Uru should be seen as ethnically separate from the Choquela, despite the conceptual overlaps they hold (Bouysse-Cassagne Reference Bouysse-Cassagne1987, Reference Bouysse-Cassagne1988; Bouysse-Cassagne & Harris Reference Bouysse-Cassagne, Harris, Bouysse-Cassagne, Harris and Platt1987; Wachtel Reference Wachtel2001). Such distinction is feasible given the sixteenth-century accounts by de Matienzo, which document that “[t]he others who call themselves chuquilas are hunters: there are few of these, and those who are they only understand how to kill feral cattle and worship. They are sorcerers, and … never see Spaniards” (de Matienzo Reference de Matienzo1967: 276, author’s translation).

Though chronicles such as de Matienzo’s describe a clear distinction between the Choquela and the Uru, the limited information available on these groups suggests that such differentiation applies only to Chucuito (a colonial province in Peru). Moreover, after the sixteenth century AD, the term Uru may have superseded Choquela, as the former was more generic and came to refer more broadly to foraging populations. The label appears to have been so inclusive that it was often used to denote linguistic categories (Bouysse-Cassagne Reference Bouysse-Cassagne1987: 47) or to describe people who paid specific forms of tribute related to hunting and fishing (Julien Reference Julien1987). Regardless of how these groups were distinguished, in the high-altitude areas where vicuñas live—between 3164 and 4857masl (González Reference González2020: 52)—there were historically foraging populations who specialised in hunting this camelid and other prey using the chacu funnel-shaped traps.

Geospatial analysis and remote sensing

The archaeological sites included in this study derive from three main sources: academic publications, grey literature and remote sensing. For the first two sources, co-ordinates were transferred to an ArcGIS Pro geodatabase, where each site was catalogued with relevant contextual information. In cases where geolocation was unavailable, place descriptions and toponyms were used, supplemented by site identification through freely accessible satellite imagery, primarily sourced from Google Earth and Bing. These images were integrated into a GIS environment within ArcGIS Pro using application programming interface keys.

A remote-sensing survey also made use of these freely available satellite imagery sources for a systematic search. A grid of 500 × 500m squares was deployed across the polygon representing the Camarones basin. This was used as a guide to systematically cover the entire study area, using a zoom level appropriate to the resolution of the available imagery. In 2018, the resolution spanned between 30 and 50cm/pixel. In these images, architectural features and large-scale or linear elements such as geoglyphs or pathways are typically the only structures visible. To ensure accurate identifications, known sites across all geomorphic floors of the Western Valleys were first examined using the imagery. This preliminary step served to familiarise the image interpreter with the appearance of archaeological sites in satellite imagery and to refine the criteria used for their identification (Snyder & Haas Reference Snyder and Haas2024).

Newly identified archaeological sites were initially recorded as points within the geodatabase, with an associated confidence level of low, medium or high. Preliminary information, including the type and estimated size of the site, was also documented. To ensure analytical rigour and minimise false positives, only records classified as having high confidence were included in further analyses. For the selected high-confidence records, additional data were examined, including the number of features per site and the geometry of any discernible structures.

Given the broad chronological range of the archaeological sites identified in the Camarones basin—spanning the entire pre-Hispanic sequence and extending into the colonial era—the range of sites observed is equally diverse. This diversity is further enhanced by the various sources of information consulted, which involve different identification techniques, including ground-level directed surveys, excavations and remote sensing (Oyaneder Reference Oyaneder2022). While this study is informed by an extensive bibliographic review, the new data presented here are derived exclusively from remote-sensing analysis.

Distribution of evidence related to macro groups

In total, 1622 archaeological sites were identified in the Camarones Basin. Of these, 12 per cent are from grey literature, 15 per cent from academic publications and 73 per cent from the remote-sensing approach. Within this diverse dataset, 790 records (49%) correspond to small-scale settlements, 58 (4%) to large settlements and 76 (5%) to hunting traps, the largest quantity of this latter category thus far identified in the Andes. Small- and large-scale settlements, as well as hunting traps, were analysed independently using the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) tool in ArcGIS Pro. This tool identifies clusters based on a defined distance between points (Birant & Kut Reference Birant and Kut2007), which in this case was set at 5km (Figure 3).

Figure 3. a) Macro-groups and above-ground vegetation distribution during dry and humid seasons; b) DBSCAN analysis of small-scale settlements; c) DBSCAN analysis of chacu traps. Basemap: hillshade derived from ASTER DEM at 30m/pixel (figure by author).

This defined distance, or epsilon, was established based on the estimated walking range over two to four hours of continuous movement without rest, within the area of highest site concentration, located between 2500 and 3800masl. A test point was defined from a hunting site at 3072masl, from which cost models based on slope were calculated to generate isochrones. These were created using the movecost R package (Alberti Reference Alberti2019), applied to a 30m/pixel ASTER DEM (advanced spaceborne thermal emission and reflection radiometer digital elevation model), incorporating Tobler’s (Reference Tobler1993) off-path hiking function, Rees’s (Reference Rees2004) slope-dependent cost function and the linear walking-time cost function (Chapa Brunet et al. Reference Chapa Brunet, García, Mayoral Herrera, Uriarte González, Bender, Evelpidou, Krek and Vassilopoulos2008). The resulting isochrones were imported into ArcGIS Pro to enable visualisation within the surrounding landscape (Figure 4), and a 5km-diameter buffer was established as a control element from the sampling point. Cost surfaces were calculated using a terrain factor of 1.67, based on conditions corresponding to bad trails and stony outcrops. Additionally, positive slope values were preliminarily multiplied by 1.99 and negative slope values by 2.31, in accordance with cognitive slope cost estimations (Pingel Reference Pingel2010). This adjustment reflects the rugged topography at these elevations, which includes ravines, abrupt drops, steep slopes and uneven terrain.

Figure 4. Projected 5km radius from central point at 3072masl, showing slope-dependent isochrones in one-hour intervals and the distribution of surrounding macro-group archaeological evidence (figure by author, basemap: Esri World Imagery).

Small-scale settlements

Small-scale settlements—clusters of up to nine standing structures and enclosures (Oyaneder Reference Oyaneder2022: 274)—are often viewed as ephemeral sites and thus marginal in studies of the Western Valleys. Recent research has begun to reposition them within broader socio-economic frameworks. In the Azapa Basin, satellite imagery revealed 560 archaeological sites, 64 per cent of which are small-scale and interpreted through pottery typology as part of Late Intermediate and Late period settlement systems (Saintenoy et al. Reference Saintenoy, Ajata, Romero and Sepúlveda2017). In the southern Atacama Desert and Salt Puna, the internodal approach has focused on sites situated between central nodes (Berenguer & Pimentel Reference Berenguer and Pimentel2017), with a particular emphasis on agropastoral socioeconomic structures. These sites correspond closely to the Andean concept of estancias: small, often seasonal settlements above 2500masl used by Uru-Chipaya and Aymara populations as pastoral outposts (Bouysse-Cassagne Reference Bouysse-Cassagne1987: 210; Wachtel Reference Wachtel2001: 68).

In the Camarones Basin, many small-scale sites remain undated due to the lack of excavations (Figure 5). Surface collections at 10 locations have yielded a Middle Archaic projectile point (6000–4000 BC) alongside Late Intermediate, Late and colonial-period decorated pottery (García & Ajata Reference García and Ajata2016; Oyaneder Reference Oyaneder2022). Local muleteers identify several sites, particularly those in the upper reaches of the basin, as majadas—seasonal pastoral residences occupied after the summer rains, typically between March and April, to take advantage of newly sprouted pasture (García & Ajata Reference García and Ajata2016: 240).

Figure 5. Small-scale settlements in the Camarones Basin: a) PhICAM 947; b) PhICAM 1414; c & d) PhICAM 1287 (figure by author, basemap: Google Earth Pro).

The distribution of small-scale architectural sites in Camarones and their correlation with seasonal changes in vegetation cover indicate a marked preference for upland zones, where low-scale stone-wall constructions are concentrated (Figure 6d). This pattern intersects with the location of chacu, which, while often overlapping with small-scale sites, are more densely distributed below 4000masl. All upland small-scale settlements fall within a single cluster under a 5km epsilon, suggesting a coherent, tethered settlement pattern oriented towards hunting.

These features suggest a landscape that reflects an accumulation of occupations over time, possibly within a broad temporal range. Moreover, the extended yet not concentrated distribution is indicative of highly mobile populations, potentially shared by culturally distinct groups. Although camelid herding is yet to be confirmed, the presence of traps and their territorial alignment with vicuña, guanaco (Lama guanicoe) and taruca (Hippocamelus antisensis) habitats support an interpretation of sustained ungulate hunting.

Hunting traps

The chacu are large installations, interpreted as traps for collectively hunting wild ungulates, consisting of funnel-shaped dry-stone walls laid out on slopes with gullies. In Camarones, 76 hunting structures were identified through a satellite imagery-based survey (Oyaneder Reference Oyaneder2022). However, a study on desert kites (another term for funnel-shaped hunting traps) on a global scale was published in parallel, reporting 102 traps in an area covering part of the Camarones Basin and other basins to the south and east (Barge et al. Reference Barge, Wazizeh, Brochier, Crassard, Régagnon and Noûs2020). These results confirm a parallelism with other arid regions of the world, where the use of funnel-shaped traps to hunt ungulates has also been documented. These traps characteristically establish vertical drops between the antennae of the trap and the enclosure.

Across the Andes, chacu traps have received limited archaeological attention, and there is little information available from excavations and surveys (Aguilar Meza Reference Aguilar Meza2018; Barnes Reference Barnes2024). As a result, the chronology of their use is uncertain. Nonetheless, colonial accounts broadly agree that chacu were employed by the Inka during the Late Period and their use continued into colonial times (Yacobaccio Reference Yacobaccio and Gordon2009). Rock art offers a complementary perspective; hunting scenes involving chacu depicted in the naturalistic style date to as early as 4000 BC, with later examples from the Formative Period (Dudognon & Sepúlveda Reference Dudognon and Sepúlveda2018). Anthropology and ethnohistory further enrich archaeological interpretations by linking these traps to hunting groups such as the Chacu haque (Bertonio Reference Bertonio1879: 68) and identifying the Choquela as key operators of chacu in vicuña capture (Millones & Schaedel Reference Millones and Schaedel1980; Cuentas Reference Cuentas1982).

The trapping function of these sites is reinforced by the distribution of chacu identified in the Camarones Basin (Figures 6 & 7). The presence of chacu between 2800 and 4200masl closely aligns with the natural habitat of vicuña, guanaco and Andean deer. In other arid regions of the world, traps with comparable morphological features—specifically long antennae converging into enclosures—have primarily been used to capture fast and elusive animals such as gazelles using roundup techniques (Crassard et al. Reference Crassard, Abu-Azizeh, Barge, Brochier, Chahoud and Régagnon2022). The herding behaviour of the fleet-footed Andean ungulates likewise makes trapping an effective hunting strategy for this prey, supporting the interpretation of chacu as hunting installations.

Figure 6. a) Scatter plot of total structures and site area, distinguishing small-scale (≤9 structures) from large (≥10 structures) settlements; b) chacu arm lengths (dominant/secondary), with circles sized by area of end corral; c) kernel density plot of altitude distributions for chacu sites (black), with modelled wet (green) and dry (red) seasons; d) equivalent kernel density plot for small-scale settlements (blue) (figure by author).

Figure 7. Examples of chacu traps found in the Camarones Basin. Note the drop found in the trap shown in the middle (figure by author).

In terms of construction, the chacu in the Camarones Basin display a high degree of consistency: 90 per cent comprise two antennae and a final enclosure, while the remaining examples include variations with up to four antennae, double enclosures or open-ended forms lacking a final corral. Notably, these traps are consistently located on steep slopes, often positioned over gullies or small watercourses running through the centre of the structure. Each trap typically features one dominant and one secondary antenna (Figure 6b), averaging 181m and 71m in length, respectively. The final enclosures vary in shape—circular, quadrangular or irregular—largely following the terrain rather than a standardised plan, with an average area of 95m².

Tethered settlement pattern in the highlands

The chacu and small-scale settlements described above suggest that a range of human groups occupied this landscape from at least 6000 BC to AD 1700. These groups moved strategically across the highlands, tethered primarily to hunting resources, particularly vicuña. The evidence indicates overlapping lifeways, combining hunting-gathering with, possibly, pastoralist-hunting practices. Foragers specialising in vicuña capture and mobile pastoralist-hunters may have been contemporaneous at times, likely requiring negotiated access to shared hunting territories. Such arrangements may have resembled the collective management of wild animals by ranching communities, involving the division of herds among family groups (e.g. Ingold Reference Ingold2007). Within this framework, some small-scale settlements in Camarones may have served as ranch-like outposts embedded in a tethered landscape of seasonal mobility.

It is therefore plausible that during the Late Intermediate and Late periods, and into the colonial period, Aymara curacas (chiefs) engaged in a form of vicuña ranching or claimed territorial control over specific hunting zones through ranching vicuña (e.g. Yacobaccio Reference Yacobaccio and Gordon2009). Simultaneously, other groups may have operated within the same landscape, actively hunting in areas overseen by curacas. The clustered yet broad distribution of small-scale settlements across the highlands strengthens this interpretation. Additionally, the seasonal variability in vegetation cover would have supported both domestic herds and wild ungulates, reinforcing the dual economic use of these highland spaces as tethered landscapes structured by both ecological and social logics.