Antiquity Vol 81 No 312 June 2007
A recently (July 2006) excavated prehistoric dwelling in far northern Peru that dates to the fifth millennium BC is one of the earliest examples of human architecture known from Andean South America. Discovered at the site of El Porvenir in the archaeologically under-studied Department of Tumbes, Peru, archaeological features consist of a relatively substantial structure 5m in diameter built before 4700–4220 BC. The El Porvenir structure is thus some 800–1000 years older than the earliest houses associated with the better-known Valdivia culture of south-western Ecuador. This discovery raises important questions about the antiquity and processes of sedentary village life in the prehispanic Andes.
El Porvenir: recent archaeological data
The site of El Porvenir is located near the modern community of the same name, in the Province of Zarumilla, Department of Tumbes, on the Peruvian-Ecuadorian border (Figure 1). El Porvenir sits on a river terrace on the southern bank of the Zarumilla River, a seasonal drainage that drains the northern slopes of the Cerro de Amotape and empties into a network of estuaries and mangrove swamps that front the Pacific Ocean.
The prehistoric site of El Porvenir consists of six mounds surrounding an open plaza, and covers an area of approximately 300 X 100m (Vilchez et al. 2007). The mounds are artificial constructions ranging in area from 160m2 (Mound 5) to 650m2 (Mound 4) with a maximum elevation of 1.6m above current ground level. Several of these mounds have been partially or completely destroyed by modern constructions since they were first recorded in 1996, but Mounds 1 and 2 are intact (Moore et al. 1996). Mounds 1 and 2 and the plaza area between them were excavated in 2006. The mound excavations were directed to the discovery and definition of the remains of superimposed floors and associated architectural features, using standard archaeological excavation techniques (hand tools, field screening with 5mm and 3mm mesh) in natural and cultural stratigraphic layers.
In Mound 1 three floors were encountered, while six prehispanic floors (and one recent floor) were found in Mound 2 (Figure 2). The El Porvenir floors were associated with three different construction methods (Table 1). The uppermost prehistoric floors in Mounds 1 (Floor 2) and 2 (Floor 1) were associated with rectilinear structures incorporating wattle and daub walls, indicated by postmolds and large fragments of fire-reddened daub with cane impressions. In lower strata, there was evidence of elliptical structures associated with well-made, thick (10–15cm) floors of grey (10 YR5/1) clay, slathered in place to form a pavement (Mound 1, Floors 3, 4, 5 and 7; Mound 2, Floors 2 and 3). Postmold patterns indicate the structures were built from paired, upright roof-posts that probably supported sloping rafter posts. Dense concentrations of fine gray ash suggests the roofing was made from some type of vegetal thatching, and the well-preserved clay floors show no evidence of erosion indicating the roofing was sufficiently thick to shed rainfall.
The earliest floor excavated at El Porvenir is Mound 1, Floor 6 (Figures 3, 4). Postmolds associated with Floor 6 indicate a relatively substantial circular structure made from paired uprights like the later structures but without the thick clay floors. Approximately 25 per cent of the structure was uncovered, suggesting a building of 18-20m2. Floor 6 was covered by a dense midden of oyster shell 30–34cm thick. This distinctive stratum consisting of more shell than soil and overwhelming comprised of Ostrea (97 per cent per weight) was encountered elsewhere at El Porvenir—in Mound 2 and in the Plaza excavations.
Charcoal and shell samples were collected for radiometric assays to establish the absolute dates of the occupations (Table 2). Shell samples were adjusted for local reservoir effect, and all dates were calibrated using the CALIB 5.0 program by Beta Analytic Inc. Radiocarbon dates indicate that most of the floors in Mound 1 date from 1280–730 cal BC, with many of the dates clustering around 940–790 cal BC. Based on this, the elliptical structures with well-made clay floors are thought to date to the tenth to eighth centuries BC (Mound 1, Floors 3, 4, 5 and 7; Mound 2, Floors 2 and 3), while the rectangular wattle and daub constructions were constructed subsequently at circa 790–400 BC.
Floor 6, however, is significantly older than these constructions, as it underlies the oyster layer that dates to 4710 – 4220 cal BC. It is important to note that while postholes from later constructions had cut into the oyster stratum in Mound 1, the dense layer was otherwise undisturbed and intact. In the profile intrusive features were readily visible as were intact deposits of oyster shells (Figure 5). Three radiocarbon samples were collected from the oyster stratum in different portions of the site: Mound 2, Unit 8, Level 5 at 102cm (BETA:222671: 4770–4490 cal BC), Plaza Operation 2, Level 4, 4370-4220 cal BC) as well as in Mound 1, Unit 8/5, Level 6 (4710-4220 BC). These dates consistently place the oyster deposit as dating to the first half of the fifth millennium BC. Floor 6 underlies this stratum and therefore was constructed before this date.
Given the lack of fieldwork in the Department of Tumbes, the closest archaeological analogies to the El Porvenir structure come from excavations in Archaic and Formative period sites in south-western Ecuador, particularly of the well documented Valdivia tradition (Lathrap et al. 1975; Marcos 1978; 2003; Damp 1984; Zeidler 1984; Staller 1994; 2001; Raymond 2003). On the Santa Elena Peninsula, Stothert uncovered evidence of a relatively small structure 1.5m in diameter associated with the Las Vegas period site of OGSE-80 with a hearth feature dated to 8920+120 (TX-4460), a shell sample that calibrates at two s to 8305-7654 cal BC (Stothert 1988). At Real Alto, Damp uncovered several structures associated with the earliest phases of Formative Valdivia, relatively small 4.5 X 3.0m elliptical structures (1984). A radiocarbon date of 5495+200 RCYBP (GX-5267), calibrated at two sigmas to 4800–3800 BC, was interpreted by Damp as predating the actual dwelling.
In Middle (3100–1900 BC) and Late Valdivia, houses increased significantly in size and substance, elliptical constructions of 12 X 8m made from dozens of relatively large posts (Zeidler 1984). As Raymond (2003: 50) has characterised this shift, 'Together with the overall growth in size of the settlement, there was a change in the size and permanence of [Valdivia] houses…. Early Valdivia houses, as evidenced at Loma Alta and Real Alto…, were only slightly larger than the Vegas shelters and no less flimsy. The Middle Valdivia houses at Real Alto, however, are substantial structures with floor areas three to five times those of Early Valdivia'. According to Marcos (2003), this shift occurred at approximately 3000 BC.
The results from El Porvenir suggest that relatively large, substantial dwellings were constructed in the Zarumilla region some 1000 years before that transformation occurred in south-western Ecuador. The antiquity and substantial construction of the El Porvenir structure is not predicted by established models of the development of sedentism in the equatorial Andes. Rather, El Porvenir contributes to an emerging body of archaeological data suggesting that the Archaic/Formative transition was a complex mosaic of which Valdivia is the best-documented manifestation, but certainly not the only one and perhaps not the earliest. For example, excavations at Santa Ana-La Florida, in humid montaña of the eastern slope of the Ecuadorian Andes, have documented a small ceremonial center and funerary structures dating to the third millennium BC, and thus contemporary, yet distinct from the Valdivia patterns (Valdez et al. 2005). As Raymond (2003: 35) has observed, research on the Santa Elena Peninsula into the Las Vegas and Valdivia traditions 'has been the vantage point from which the Ecuadorian Formative has been interpreted' a 'peephole' into the past characterised by 'the narrow regional bias of the [Santa Elena] peninsula'. The investigations at El Porvenir and other prehistoric sites in the septentrional Andes are amplifying the vista on the development of sedentism in north-western South America.
Tables Click on the table name to download.
Table 1. Prehistoric architectural patterns at El Porvenir, Department of Tumbes, Peru.
Table 2. 14C dates from El Porvenir, Department of Tumbes.
The research at El Porvenir was conducted as part of the Proyecto Arqueológico Tumbes, funded by the National Science Foundation (Grant: 0549454) and authorised by the Instituto Nacional de Cultura (Resolución Directoral Nacional No. 783/INC); I thank these institutions for their support. I wish to thank my co-director, Carolina Vilchez, and colleagues, Bernardino Olaya and Eva Pajuelo for their multi-faceted contributions to the project. I appreciate the efforts of Milagros Obregoso, Catherine Lopez, Guadelupe Gonzalez and Paul Flores who participated in the excavations. I also wish to thank my colleagues John Staller, Andrew Stewart, Janine Gasco, and Matthew Deslauriers who read earlier drafts of this report. All errors or omissions are my responsibility alone. Finally, I wish to thank the people of El Porvenir for their hospitality and participation in the project.