Geochemical soil survey as a proxy for ancient human activity at Sagalassos (SW-Turkey)

Ancient human activities such as metal smelting, food consumption and refuse dumping have resulted in the addition of a wide range of organic and inorganic materials to soils at occupation areas, thereby altering their geochemical composition. These additions can be detected up to the present day, and hence provide a tool to address archaeological questions. While geochemical soil studies have frequently been applied on exposed floor layers at archaeological sites, they have rarely been considered within the research domain of archaeological prospection, where they could serve as a means for providing information on the buried archaeological record. This thesis assesses the potential of this approach on several sites in the Roman to Byzantine city of Sagalassos (SW-Turkey), which offers an excellent test location because of its extended archaeological history, varying physical landscape and well-developed scientific framework. As several studies have recommended the integration of different survey methods in order to obtain a more objective understanding of the history of buried sites, special emphasis is put on the use of a multi-analytical approach.Three main research aims were put forward in this thesis. We evaluated how the chemical composition of soils on unexcavated sites at Sagalassos was altered by ancient human activities (1). Also, we studied to which extent soil geochemical data are useful in providing information on the former functionality of (zones within) archaeological sites at Sagalassos (2). Finally, we examined the potential of integrating soil geochemical survey with other survey techniques (3). To answer these research questions, we performed geochemical surveys on four study areas, each with a different historical and geological context. The first site subjected to a soil survey was Catal Oluk, which is an isolated rural site, seemingly not directly influenced by pollution inputs from the city. Next, samples were taken at the Eastern Suburbium, an area situated in the east of the ancient city of Sagalassos. This area has witnessed a variety of intense human activities over a long period of time. Thirdly, a survey was performed in a large area west of the Colonnaded Street. Finally, samples were collected at Gökpinar, an area rich in bloated ceramic slag and with an unknown functionality at the time of geochemical soil survey.From a methodological point of view, we combined soil multi-element chemistry (ICP-OES analyses after Aqua Regia destructions) with the analysis of additional parameters, such as pH, organic matter, Pb-isotopes and mineralogy. Also, emphasis was put on developing a suitable statistical approach, which included boxplots, correlation analysis, principal component analysis, Ward s clustering, contiguity constrained clustering and parallel coordinate plots. These techniques were applied to compare chemical data with local background values, study chemical associations, define (spatial) groups and identify inter-group chemical differences. To study the potential of integrating chemical survey with other techniques, chemical data were combined with results from magnetic- and archaeological surveys in the study areas.For the site of Catal Oluk, the results revealed anomalies of K, P and Zn on a location where archaeological and geophysical data suggested the presence of ceramic producing kilns. These enrichments are thought to result from burning wood as fuel for the detected kilns; In addition, local anomalies of Co, Cr, Fe, Mg, Mn and Ni were found to reflect the working and storage of ophiolitic clays, which were employed as a raw material for ceramic production. Changes in Al, As, Ba, Ca, Na, Pb, Sr and Ti-content were attributed to the combined contribution of limestone and volcanic tuff fragments.In a first sampling grid constructed in the Eastern Suburbium, an area with ancient human enrichments of Cu, P, Zn and K was detected. This area was distinguished from natural patterns created by limestone and volcanic tuff (Al, Pb, Ba, As, Sr) and ophiolitic bedrock material (Co, Cr, Ni, Mg, Fe, Mn and V), but no specific interpretations regarding ancient functionality could be made. In a second, more detailed grid, a more detailed research question was put forward, i.e. offering an explanation for an area in which the magnetometry image was difficult to interpret. It was demonstrated that a zone characterised by high frequency magnetic anomalies and a low variogram-range was associated with chemical anomalies of Cu, K, P, Zn and potentially also Pb. This contributed to the interpretation of this zone as an area of magnetic disturbance created by occupational debris. Next, a zone with a low magnetic signal and low sill-values could be interpreted as a region where limestone bedrock is located close to the soil surface, based on the association with a cluster high in Al, Ba, As and Pb. Finally, two zones of enhanced magnetization and high range- and sill values were shown to spatially overlap with two clusters of soils containing elevated levels of Co, Cr, Mg, Mn, Ni and Ti, V and Fe respectively. By combining these chemical data with magnetic susceptibilities and X-ray diffraction patterns, it is shown that these two zones were the result of the presence of two partly different types of ophiolitic material near the surface.The grid west of the Colonnaded Street aimed to evaluate the characteristics and spatial extent of an ancient pollution signal in the surroundings of this street. Cu, P, Pb and Zn were shown to be enriched around the Colonnaded Street. While enrichments of Zn and Cu were weak or limited to a few samples, P and Pb were strongly anomalous, reaching values respectively 2 and 3 times higher than maximum background concentrations. The spatial extent of the pollution signal was different for P and Pb, with Pb reaching background values at 150 m away from the street, while P was at background levels at 450 m away from the street. Two explanations were put forward to explain this difference. The first assumes that all anthropogenic P and Pb stem from activities in and aroundthe ancient citycentre and the Colonnaded Street, and that surface runoff processes transported P to larger distances when compared to Pb due to differences in partitioning of P and Pb between the solid-liquid phases in the soil. The second option assumes that the high P concentrations in a zone west of the monumental centre are the result of direct P input in the soil, as the area functioned as a residential quarter in ancient times. This scenario hence suggests that the activities in the monumental centre and the western residential quarter were different in such a way that only those in the monumental centre resulted in Pb input in the soil. A link with metallurgical activities around the city centre can be put forward to explain this difference, although contributions by other activities cannot be excluded.The research in the Gökpinar area aimed to provide better insights in the functionality of this site. However, no enrichments of potentially anthropogenic elements were found. Observed chemical patterns in the sampling grid were ascribed to geological variations within the Gökpinar area, with ophiolitic bedrock material outcropping in the south and limestone rocks outcropping in the north of the grid. This conclusion is consistent with the results of a recent survey campaign, suggesting that the Gökpinar area mainly functioned as an access road towards the Eastern Suburbium.The potential of Pb-isotopes to distinguish between polluted and unpolluted soil samples at Sagalassos was evaluated in a separate chapter. It was concluded that soils strongly polluted due to ancient human activities may sow a limited shift in their Pb-isotopic composition. However, the isotopic data did not provide new information on potential ancient pollution sources in the study areas, and their applicability as tracers of ancient pollution in Sagalassos appears to be limited. There are several reasons for this, with the broad and overlapping isotopic fields of background data being one of the most important factors complicating interpretations.More in general, the results of this thesis show that Cu, K, P, Pb and Zn are locally enriched due to ancient activities, with P being the most reliable indicator of human activity. Inter-site differences are a consequence of changes in bedrock composition and archaeological history. When applied as a standalone technique, problems of superposition and mixing at the site level, and inconsistencies in literature data regarding specific element-functionality relations, prevent the use of geochemical data as interpretational tools on most of the sites investigated in this thesis. It is expected that this conclusion holds for many other sites influenced by a long history of human occupation. However, when combined with other survey techniques, such as geophysical prospection, soil geochemical survey is shown to have much more potential. Because geochemical information is to a certain extent diagnostic for the subsurface features creating geophysical anomalies, geochemical data are highly valuable indistinguishing geogenic and anthropogenic anomalies. This thesis emphasises the value of using an integrated multivariate data analysis approach. In addition, it is demonstrated that taking into account the spatial component of the chemical datasets in the statistical analyses significantly enhances the quality of the data analysis results.nrpages: 234status: publishe