Geoarchaeology of urban space in tropical island environments: Songo Mnara, Tanzania

Past urban settlements in tropical island environments offer particularly challenging sites for mainstream archaeology. Often associated with shallow stratigraphic sequences, archaeological sediments and soils in these sites are strongly influenced by local geology and seawater. This study discusses the advantages and challenges of developing an integrated geoarchaeological programme to examine the use of space at the Swahili stonetown of Songo Mnara Island, Tanzania. This exceptionally well preserved site, occupied for less than two centuries (C14th-16th AD), comprises a complex urban layout with stone-built houses, wattle-and-daub structures, funerary complexes, activity areas such as wells, and open areas. The programme has combined geoarchaeological (soil macro- and micromorphology, ICP-AES, pH, EC), geophysical (magnetic susceptibility) and archaeological (large excavations, test trenches, artefact distribution mapping) techniques to investigate the use of space across different contexts. Initial geoarchaeological prospection and opportunistic soil sampling have allowed framing of the island's environmental settings and archaeological deposits as well as outlining open spaces in between buildings. Subsequent research applied a systematic sampling strategy to map geochemical and artefact distributions in conjunction with context-specific soil micromorphology. The results provide a means to map out the impact of occupation across the site as well as to differentiate between open, roofed and unroofed spaces. ICP-AES results, for example, demonstrate that measurements of Ca, Mg, P, S and Sr levels can help discriminate occupation/activity areas in tropical island environments. They also indicate that the depletion of certain elements (e.g. Na, K, and Ni) should be considered as a means of differentiating between roofed and unroofed spaces. The combination of different methodologies demonstrates the importance of addressing discrepancies as well as correlations between multiple datasets for deciphering features within urban spaces in tropical environments and interpreting ancient activities that occurred within them.