CHRONOLOGICAL AND REGIONAL PATTERNS

The earliest known accepted evidence of human occupation in the Amazonian floodplain comes from Pedra Pintada cave, located in Monte Alegre, on the north bank of the lower Amazon. It is a sandstone cave where excavations done in the early 1990s revealed evidence of human occupation dating back to ca. 9200 years BC (Roosevelt et al. 1996, 2002). At Pedra Pintada, the abundance of plant and small faunal remains suggest a diversified economy rather than specialized big-game hunting. The same picture emerges from the sites located in the Araracuara area, in the Caqueta river in Colombia where the abundance of palm remains also suggests a diversified, non-specialized economy dating back

To ca. 7000 years BC (Mora 2003). In the lower Jamari River, a tributary of the upper Madeira River located close to the present-day city of Porto Velho, Miller et al. (1992) have identified a long sequence that covers the whole Holocene, with the early dates going back to 6800 BC. At the Dona Stella site, in the central Amazon, ongoing excavations have found bi-facial lithic artifacts, including projectile points, possibly dating to 5700 BC (Neves 2003). In the Caraja plateau of eastern Amazonia, located between the Xingu and Tocantins Rivers, excavations done in rock shelters have also produced evidence of human occupation dating back to the early Holocene (Magalhaes 1994). The fact that these sites are located at places far away from each other - and in some cases away from the large rivers-demonstrates that different parts of the Amazonian floodplain were already occupied by the early Holocene. Also from Pedra Pintada and from the fluvial shell mound of Taperinha, located downstream from modern-day Santarem, Roosevelt has excavated grit-tempered pottery dating to the sixth and fifth millennia BC (Roosevelt 1995; Roosevelt et al. 2001, 2006). Confirmed by slightly later dates obtained for shell-tempered pottery from the Mina phase, excavated at maritime shell mounds on Atlantic coast east of the Amazon mouth (Roosevelt 1995; Simoes 1981), they are the earliest dates for ceramic production in the Americas.

Based on Lathrap’s influential “cardiac model,” which proposed that the Amazonian floodplain was continuously and densely occupied since the mid-Holocene (Lathrap 1970a, b; Lathrap and Oliver 1987), there has been a tendency to imagine demographic pressure along the floodplain as early as 3,000 years ago (Hornborg 2005; Oliver 2001). However, the available archaeological data are not supportive of this idea (Heckenberger, Neves and Petersen 1998; Neves and Petersen 2006). To the contrary, the data show that, with the exception of the already mentioned upper Madeira basin (Miller et al. 1992) and the lower Amazon (Meggers and Danon 1988; Roosevelt 1991a; Schaan 2001a, 2004), signs of human occupation during that time are quite scanty. For instance, in the central Amazon—Lathrap’s putative center of early cultural development in lowland South America—evidence of human occupation from 5700 to 500 BC is absent, despite the identification of more than 100 archaeological sites in a 900 km2 research area (Neves 2003). At Pena Roja, on the Caqueta River, despite early dates showing occupation of the area ca. 9200 years BP there is a hiatus during the mid-Holocene, interrupted only around the Christian era (Cavelier, Herrera, Morcote and Mora 1995: 27; Herrera, Bray and McEwan 1980-81; Mora 2003: 91-92, 126; Mora, Herrera, Cavelier and Rodriguez 1991). A third example comes from the Santarem area. There, after the above-mentioned occupations with early ceramics at Pedra Pintada and Taperinha dating from ca. 6000 and 5000 BC, there is also a hiatus which is only interrupted much later on, with sites of the Poco phase dating to 100 BC (Hilbert and Hilbert 1980), despite the recent discovery of occupations in the lower Tapajos with dates back to 1800 BC to 1200 BC (Gomes 2005: 231). In this area, however, there are also gaps between these early occupations and later occupations from the Christian era.

How should we interpret these hiatuses? Do they mean that the Amazon basin was scarcely occupied during the mid-Holocene? Is there a taphonomic bias resulting in the destruction or poor visibility of sites dating from this interval, possibly related to intensive soil erosion (Mora et al. 1991: 41-43)? Can these apparent hiatuses be correlated to climatic change events? Given the modest amount of research in the Amazon we are still far from answering these questions, but it is likely that all of these factors contribute to the current picture. Among the archaeologists working in the Amazon, Meggers has been the most forceful in pursuing positive correlations between climatic change—mostly supposedly related to

ENSO episodes—and gaps in the regional sequences (Meggers 1977, 1979, 1982). The problem is that, at the time, little research had been completed on Holocene climatic patterns. Data from pollen records (Absy and van der Hammen 1976; Baker et al. 2001; Behling and Hoog-hiemstra 2000; Behling and Lima da Costa 2000; Behling, Berrio and Hooghiemstra 1999; Behling, Keim, Irion and Nunes de Mello 2001; Berrio 2002; Burbridge, Mayle and Killeen 2004; Haffer 2001; van der Hammen 2001; van der Hammen and Hooghiesmstra 2000), carbon isotopes in stable organic matter (Freitas et al. 2001) and hydrology (Aalto et al. 2003) show that the mid-Holocene could have been drier than the present, with a potential impact on the distribution and frequency of plant species without, necessarily, the development of open savanna vegetation and forest refuges proposed earlier (Meggers 1977, 1979). An alternative is also to propose that climatic variation at the time was probably more significant than previously thought, with a potential impact for human occupation. In the lower Xingu river a Mina phase fluvial shell mound with ceramics dating from 2000 BC has its basal strata under water, even during the dry season, which is interpreted as evidence that it accumulated during a time of drier conditions, when the water level was lower than today (Perota and Botelho 1992). At Chambira, in the upper Amazon, a ceramic assemblage composed by vessels with restricted forms, single and double spouts and small mouth diameters dating from 1500 BC to AD 1 has been interpreted as a technology adapted to minimizing evaporation at a time of drier climatic conditions (Morales 1998).

If the still tenuous evidence for drought during the mid-Holocene holds, it is likely that many of the archaeological sites from this time are today either destroyed, under water or even buried by tons of alluvial sediment. Conversely, one needs also to consider the possibility that drought and its subsequent changes on water level and forest cover may have had direct impacts on human occupation of the Amazon during that time. Traditionally, supposed reduction of the forest into refuges during the Holocene has been interpreted as conductive to human expansion into the Amazon (Meggers 1977). The archaeological and climatic data from the mid-Holocene suggest the opposite: human occupation surged only after current tropical climatic and ecological conditions were reached about 1000 BC. From this time on, one again sees strong and unequivocal signs of settlements in the floodplains and adjacent areas, a pattern that becomes clearer and stronger after the onset of the Christian era.