摘要：It has recently been argued that pre-Columbian societies in the greater Amazon basin during the Late Holocene were subject to “adaptive cycling”. In this model, cultures practicing “intensive” land use practices, such as raised field agriculture, were vulnerable to perturbations in hydroclimate, whereas “extensive” land use patterns, such as polyculture agroforestry, are viewed as more resilient to climate change. On the basis of radiocarbon data, the relative rise and fall of late pre-Columbian cultures and their inferred patterns of land use in six regions are highlighted to exemplify this model. This paper re-examines the radiocarbon evidence marshalled in favour of adaptive cycling, demonstrating that alleged temporal patterning in these data are overwhelmingly likely due to a combination of sampling effects, lack of statistical controls, and unacknowledged uncertainties that are inherent to radiocarbon dating. The outcome of this combination of factors seriously limits the possibility of cross-referencing archaeological data with palaeo-ecological and -climatological data without controlling for these effects, undermining the central archaeological pillar in support of adaptive cycling in Amazonia. This paper illustrates examples of such mitigation measures and provides the code to replicate them. Suggestions for how to overcome the serious limitations identified in the Late Holocene radiocarbon record of Amazonia are presented in the context of ongoing debates on inferring climatic causation in archaeological and historical datasets.
