Abstract
The recovery and identification of wood charcoals from soil not associated to archaeological excavation is called pedoanthracology. Researchers in this field use different sampling methods, and only one study, in a temperate context, has compared their efficiency so far. In this paper, we compare the efficiency of three sampling methods on charcoal concentration (> 4 mm) and on taxonomic representativeness in a tropical environment for the first time. These methods are pit sampling, auger sampling and opportunistic handpicking of charcoal on tree uprooted mounds or in the soil excavated during the digging of a pit. Our results suggest that the two digging methods (pit and auger) allow to recover similar charcoal concentrations whatever the site and the depth under consideration. But as expected, the charcoal concentration of a site depends on its history (e.g. type of activity). We also show that the estimation of the minimal sampling effort required to obtain taxonomically representative assemblages, in terms of number of charcoals and of auger samplings, varied greatly depending on the site and on the presence of overrepresented taxa. Finally, our results show that auger cores were often monospecific (34 %). Nevertheless, they allowed to recover different taxa from the pit method, with 18 to 21 % of the taxa exclusive to the first method and 30 to 40 % of the taxa exclusive to the second one in the sites under consideration. Charcoals from opportunistic handpicking also allowed to improve the taxonomic diversity of the whole assemblage of a site with 5 to 15 % of the taxa being exclusive to this method. The three methods are therefore complementary and we suggest to use the three of them to obtain the best taxonomic diversity in an anthracological assemblage and to overcome the specific biases of each method. With this study, we hope that we will help tropical (pedo)anthracologists to optimize charcoal sampling in anthropogenic sites were the sampling design cannot benefit from archaeological excavations.