The unique functioning of a pre-Columbian Amazonian floodplain fishery

International audienceArchaeology provides few examples of large-scale fisheries at the frontier between catching and farming of fish. We analysed the spatial organization of earthen embankments to infer the functioning of a landscape-level pre-Columbian Amazonian fishery that was based on capture of out-migrating fish after reproduction in seasonal floodplains. Long earthen weirs cross floodplains. We showed that weirs bear successive V-shaped features (termed 'Vs' for the sake of brevity) pointing downstream for outflowing water and that ponds are associated with Vs, the V often forming the pond's downstream wall. How Vs channelled fish into ponds cannot be explained simply by hydraulics, because Vs surprisingly lack fishways, where, in other weirs, traps capture fish borne by current flowing through these gaps. We suggest that when water was still high enough to flow over the weir, out-migrating bottom-hugging fish followed current downstream into Vs. Finding deeper, slower-moving water, they remained. Receding water further concentrated fish in ponds. The pond served as the trap, and this function shaped pond design. Weir-fishing and pond-fishing are both practiced in African floodplains today. In combining the two, this pre-Columbian system appears unique in the world. Although archaeological artefacts document diverse fish-capturing gear, from spears and hooks to nets, traps and weirs 1,2 , and rock art depicts fishing scenes, including fish traps 3-5 , the large-scale ecological functioning of fisheries and its consequences for the social organisation of fishing activities in the past are often difficult to infer. Some studies have used large collections of fish remains such as bones, scales and otoliths 6 to document the species harvested and attempt to infer how fishing activities may have been organised 7,8 , using knowledge about the biology and ecology of the harvested fish species. However, for many parts of the world such studies are rare. In Amazonia, for example, despite the great productive potential of river and floodplain fisheries and the contemporary importance of inland fisheries in the region 9 , there has been little investigation of fisheries in the past. Only rarely do archaeological remains permit functional analysis of large-scale sophisticated fish-harvesting systems 10. The study by Greene et al. 11 of a weir-based estuarine fishery on Canada's Pacific coast is such an example. This fishery relied on tidal fluctuations and the fish movements associated with them. During the ebbing tide, fis