Opinions on the extent of pre-Columbian human disturbance in Amazonia have shifted from the traditional “minimal impact” towards a basin-wide “manufactured landscape”. Evidence of large-scale landscape transformations has been found primarily along the bluffs of major rivers and in drier seasonal forests. Several lake sediment records have also recorded increased human activity during the late Holocene, although some contained no historical disturbance. Modern vegetation patterns are even predicted to result from large-scale pre-Columbian disturbances (e.g. large tracts of bamboo forests in western Amazonia). Nonetheless, the extent of human impact away from rivers and marginal forests remains poorly documented.
Here, we examined whether vast tracts of forest in western Amazonia were “manufactured landscapes”, similar in scale to major archaeological sites. We also explored whether specific parameters can be used to predict patterns of pre-Columbian disturbance.
To look for evidence of human activity (pottery sherds, terra preta, fire, and agriculture), ~500 terrestrial soil cores were collected throughout western and central Amazonia. Sampling included sites at long-term forest research stations, around lakes with sediment records containing evidence of pre-Columbian occupations, bamboo and adjacent terra firme forests, and at various distances from major rivers. Charcoal abundances were analyzed from all cores, and phytoliths, which can identify cultivars or large forest clearances, from selected cores. Soil charcoal and phytoliths were directly 14C dated to provide temporal frameworks of disturbances.
No soil core collected contained evidence of terra preta or artifacts. Many sites had no sign of disturbance, and none contained evidence of landscape-scale forest transformations. Of the long-term forest plots surveyed, fire frequency increased from 1000 – 4000 cal yr BP at Los Amigos but with no signal of agriculture, and Cocha Cashu experienced little to no historical disturbance. Charcoal presence was not significantly higher in the bamboo forests than in adjacent terra firme forests at any location. Localized disturbances with low-impact agriculture were found around lakes with known occupations. Analysis is still in progress for sites extending from river bluffs, but preliminary data suggest that historical disturbances were less frequent at distances over 5 km from the river bluffs.
Our results suggest that Amazonia was not a manufactured landscape across the basin, but instead disturbances were patchy, never homogenous, and usually located near river bluffs or lakes. Future efforts are to create Bayesian models predicting patterns of pre-Columbian ecosystem modifications, which can then be refined with further field testing.