Assessing natural versus anthropogenic horizons through deposit modelling strategies at the Medieval Site of Santa Margarida (Martorell, Barcelona, Spain)

Geophysical surveys have been one of the most useful tools for non‐invasive archaeological field assessment. Whereas they have been regularly put into practice at the archaeological site of Santa Margarida (Martorell, Barcelona, Spain), the particular features of the site required a more specific strategy to identify the contact interface between anthropogenic and natural deposits and, therefore, to conclude the archaeological excavation. As the archaeological site is settled on Quaternary alluvia at the confluence of two rivers, and in the absence of built structures in some areas, it was extremely difficult to distinguish between anthropic layers and those silty or sandy natural deposits during fieldwork. Extensive soil survey by means of borehole logging, full‐coring, and Dynamic Penetrometer Super Heavy testing within this site has proved to be a useful strategy to identify the geological horizon and to map the original relief of past landscape before excavation tasks. This enabled us to plan our fieldwork better and to interpret particular features of the site. This paper aims at summarising the results of recent surveys and discussing the potential of this approach for the study of rural settlements and agrarian landscapes in the Middle Age.


| INTRODUCTION
The geoarchaeological approach, broadly understood as the application of geosciences to solve research problems in archaeology (Kokinou, 2015, p. 209;Pollard, 1999, p. 7;Wilson, 2011, p. 2), is a long-lasting consolidated practice in the archaeological domain for multiple research purposes. In the last 50 years, the finding and location of archaeological sites have considerably benefited from the advances in geophysical survey techniques, which has enhanced the design and planning of archaeological excavation in a more costeffective way. Non-intrusive methods, such as ground-penetrating radar (GPR), electromagnetic induction methods, electrical resistance techniques, and magnetic or acoustic assessment, among some other less frequent practices (Sarris et al., 2018, pp. 224-226), have been crucial for site exploration and mapping. The archaeological project of Santa Margarida (Martorell, Barcelona, Spain) has pioneered the geoarchaeological approach in Catalonia since it began in the 1980s (Farreny et al., 2012, p. 368;Mauri, 1988, pp. 8-10).
The site of Santa Margarida is settled on a Quaternary alluvial plain at the confluence of two rivers. The most evident vestige is a Romanesque temple built upon the structure of an Early Christian church and surrounded by an extensive necropolis as well as habitat and storage structures. After more than 40 years of archaeological fieldwork, excavation tasks apparently concluded in some areas when artefact-lacking clayey layers were reached and attributed to the natural alluvial deposits. However, more recent excavationsplanned according to the GPR survey results (Sala et al., 2010(Sala et al., & 2011-revealed that some of those deposits were, in fact, abandonment layers covering earlier occupation horizons. Certainly, in the absence of built structures or artefacts in some areas, it was extremely difficult to distinguish between anthropogenic layers and those silty or sandy natural deposits during fieldwork. The issue of archaeological "non-visibility" (Campana, 2009) became then a major challenge for archaeological investigation at the site and fostered the search for other strategies to determine whether the excavation tasks could be concluded or not and to explore the archaeological potential of adjacent areas.
The main research problem to solve was the correct identification of the natural horizon after removing the archaeological deposits and artificial ground. Distinguishing this geological interface between clayey and silty layers with extremely feeble human evidence (or none at all) required a more precise assessment of the geological environment features, for which some proposals have been made in the past. Existing literature on this topic (Artz, 2006;Carey et al., 2019;David et al., 2008, pp. 498-499;Kokinou, 2015, pp. 210-211;Layzell & Mandel, 2019) demonstrates how a combined approach using geophysical techniques in conjunction with a borehole sampling can provide a cost-effective methodology for modelling subsurface stratigraphy (Bates & Bates, 2000, p. 845).
Extensive soil survey by means of combined borehole logging, Dynamic Penetrometer Super Heavy (DPSH) testing, and undisturbed sample full coring and analysis within this site has proved to be a useful strategy to identify the geological horizon and to map the original relief of past landscape before excavation tasks in new trenches. This enabled us to plan our fieldwork better and to interpret particular features of the site. This article aims at summarizing the results of recent surveys and discussing the potential of this approach for the study of rural settlements and agrarian landscapes in the Middle Age.
In fact, by means of analysing the results obtained at Santa Margarida, we aim at offering some thoughts about the relation between environmental features and settlement history, and about the way in which geological conditions at the site are concomitant with the post-depositional affectation of archaeological vestiges as well. By sharing our study case results with a broad audience throughout this paper and discussing the pros and cons of our strategy, we might help other archaeologists in need of determining accurately the natural horizon distinction at their sites from a geoarchaeological approach. The first permanent settlement is evidenced by a 5th-century-AD Early-Christian church with a tripartite apse and its extensive necropolis surrounding the temple (Navarro & Mauri, 1993).
A horseshoe-arched entrance on the South wall leads to an annexe hall building, probably used for funerary purposes. Another attached building on the West wall might have been used as a baptisterium.
These annexe buildings did not last for a long time, as the burials of the Early Christian Necropolis (5th-7th centuries AD) were arranged in order around the temple once they had already been demolished ( Figure 1d). During the period between the 6th and the 12th centuries, the site experienced a great transformation in accordance to the origin and early development of the medieval village of Martorell and the successive refurbishment of the church, finally transformed into a Romanesque building (Navarro & Mauri, 1993, p. 342;Travé et al., 2019, p. 186).
During the second half of the 11th century AD, while the Early Christian church was still in use, some medieval cob-built habitat structures were placed in the nearby area. Within an adjacent radius of 30 ft around the church, peasants were protected against the feudal landlords' violence during the 10th and the 11th centuries.
This so-called sacraria or sacred area provided a safer place to live, and a great number of silos for grain storage were discovered together with the houses. At this moment, the neighbouring monastery of Sant Genís de Rocafort was built on the top of a hill at the northwestern slope of the Ataix Mountains, and Santa Margarida became the parish church depending on the monastery (Figure 1c).
At the end of the 12th century, the parish was in a blossoming period and the Early Christian building was partially demolished to build up a new Romanesque church (Figure 2a,b). From this moment on, the archaeological sites of Santa Margarida and Sant Genís would share a common history.
Archaeological research had focused on the church building and its closest 750-m 2 -wide area until we enlarged the excavation trench in 2012. Since then, extensive excavation of a 1500-m 2 -wide area has allowed us to gain a better insight into the habitat structures around the church and their chronology, but it has also raised new research questions regarding the landscape structure and layout.
Preliminary results show alternative periods of habitat and funerary uses (Travé et al., 2019, pp. 184-185). There are up to three building phases, the latest one related to a piping channel, and some of these buildings reuse those of the earlier period, slightly changing their F I G U R E 1 Location map of the archaeological sites mentioned in this paper (a-c) and layout plan of the Early Christian and Romanesque Church of Santa Margarida (d) | 513 orientation ( Figure 2c). Plenty of silos pierce the occupation and burial horizons and provide a chronological reference due to the abundance of pottery sherds within their filling deposits (Navarro & Mauri, 1997, p. 95;Travé et al., 2019, pp. 186-187).
The site was occupied and exploited as a farm until the first half of the 20th century. This continuous occupation provides a complete diachronic sequence of burial practices for the late antique, medieval, modern and contemporary eras, but it also affects the visibility of remains from the earliest periods. Certainly, many researchers in Western Europe and the Mediterranean (Francovich, 2004;Quirós-Castillo, 2011, p. 380;Reynolds, 2003;Vigil-Escalera, 2010, p. 1) have addressed the problem of "non-visibility" of archaeological vestiges in specific contexts. The study of the early medieval centuries in Spain has been one of those scenario. As a matter of fact, there is a strong relationship between visibility and continuity of settlement, as the enduring occupation of the same location interrupts ancient deposits and features, together with agricultural activity (Schofield, 1991), which also has an impact on subsurface deposits (Campana, 2009, pp. 8-10).
The geological context of this site has also challenged the ar-

| Survey areas and deposit modelling methods
The archaeological site is part of the municipal land property, close to the local cemetery and enclosed by a fence. The access path to Santa Margarida runs in parallel to the motorway leading to the urban nucleus and is perpendicular to the path to the cemetery, which leads to a local road to the town ( Figure 4). Geophysical GPR surveys carried out in the past were always limited to the archaeological area inside the fence and have regularly provided valuable information about the subsurface features (Sala et al., 2011), later corroborated by the archaeological excavation (Travé et al., 2019, p. 182).
F I G U R E 3 Geological map of the site's environment with main orographic accidents labelled As stated above, as long as the fieldwork progressed, a yellowish artefact-free clayey layer was uncovered in some of the areas, which initially led us to conclude that the archaeological tasks were completed there. Before the trench enlargement in 2012, a geological borehole logging survey was carried out inside the site's fence to determine the archaeological potential of the area by identifying the anthropogenic layers and distinguishing the interface's relief between the packages of anthropogenic sediment and the original alluvium. In 2011, a sum of twelve 10-cm-in-diametre boreholes was logged and a selection of core samples was obtained for further lab examination, as detailed in Table 1. Those areas where GPR-detected anomalies were more imprecise or confusing were selected for logging, and selected core samples were extracted as cylindrical sections by drilling with a ring-lined barrel sampler (ASTM D3550, 2017). Archaeological investigation has demonstrated that the archaeological vestiges of habitat structures continue to the east, out of the municipal land property, but we did not know what happened in the subsurface to the west. Enclosed by the paths leading to the town, there is a 12,000-m 2 area of pine forest and ground vegetation.
The area was presumed to have archaeological potential; but being densely wooded, the use of common geophysical non-intrusive methods probably would not have led to reliable data (Sarris et al., 2018, pp. 231-232). Deposit modelling (Carey et al., 2017(Carey et al., , 2019 carried out in 2011 produced successful results, which led us to plan a more extensive survey outside the site's limit, and 25 same-size boreholes were logged in the forest. Boreholes from inside the archaeological site were correlatively numbered from 1 to 12. Boreholes in the external surveyed area were also numbered correlatively from 1 to 25. Subscript labels reading "in" or "out" attached to the borehole number indicated the survey area of each one of them. Visual description of sediments and soils was carried out constantly while drilling in both areas, which provided direct information about the subsurface sediment architecture and stratigraphy (Artz, 2006, p. 20;Carey et al., 2019, p. 498). Full coring was performed at different depths and 59 nonaltered 60-cm-long cores were sampled and lab-analysed.
These analyses included sample sieving and plasticity index calculation for the fraction passing 425 µm sieve. Obtained results allowed us to classify samples within standard geotechnical soil groups (Burmister, 1949;García-Gaines & Frankenstein, 2015)

T A B L E 2 Highway Research Board (HRB) Classification of Soil (HRB, 1952)
Group classification      usually with values >10%) show a strong variability as well. Table 4 summarises the main features of the stratigraphic layers described above, including basic classification data and a description of each unit.  (Table 5). Higher variability in the anthropogenic deposits, as indicated by values of standard deviation and covariance, is due to the fact that several layers were penetrated during the survey while only the lower quaternary terrace (Ql1) was affected.

| Soil penetration resistance
Considering that raw data represented in Figure 8a  showing an equivalent resistance of the subsurface.

| DISCUSSION
Interaction between archaeology and geology has been fructiferous for a long time and has provided valuable data regarding the exploration of new sites or areas with archaeological potential, the study of environments and landscapes and their human transformation, and the exploitation of natural resources or raw materials for different purposes (Pollard, 1999, p. 8). The geoarchaeological approach in Santa Margarida has enabled us to map the features and relief of ancient natural landscape ( Figure 9) and to identify significant changes that occurred in the past, mostly due to human agency. Human interaction with the geosphere is a strong factor in landscape transformation (Wilson, 2011, pp. 4-5), particularly from the Middle Age on (Price et al., 2011(Price et al., , pp. 1060(Price et al., -1061. informs about the refurbishment tasks carried out in the area (Baucells, 2007, p. 79). Probably at this time, the site experienced its first significant transformation, although the paleorelief still preserved much of its former appearance. The identified water flow surrounding the church (Figure 10)  | 525 quarry waste (Rosenbaum et al., 2003, p. 408) of the motorway's building work. This surface layer is usually removed by machine under archaeological supervision. It is evenly spread throughout the entire area and is likely to cover most of the external surveyed extension as well, as suggested by some of the dynamic probe tests.
High resistance peaks of surveys 4 out , 10 out , 19 out , and 22 out were related to higher percentages of CaCO 3 a 57-58 m deep. The presence of CaCO 3 -rich materials usually leads to excessively high DPSH values (MacRobert et al., 2011, p. 53  probing. Actually, we are aware of the geotechnical criticism that dynamic tests received in some specialised literature on this topic, usually because of presumed poor repeatability and accuracy (Broms & Flodin, 1988;MacRobert, 2017, p. 11). As archaeologists ourselves, we took into account a few considerations to mitigate the weaknesses of DPSH: first, that dynamic probe is an economic and easy test that might be practical for archaeological purposes, which were not related to engineering assessment of soil suitability for other more compromising activities. It should be noted that the parameters measured during geotechnical investigation differ significantly from the parameters required by the geoarchaeologists (Bates & Bates, 2000, p. 847). Second, that borehole logs and subsequent deposit modelling were the main criteria to assess the archaeological potential of the site and, hence, those quantitative and qualitative data allowed for a more comprehensive discussion of soil resistance trends and its interpretation, as considered above and as also required in geotechnical approaches.
Third, that soil variability is a complex attribute resulting from many disparate sources of uncertainties, particularly when soil data from different geologic units are mixed (Phoon & Kulhawy, 1999, p. 614). That is why a statistical examination of DPSH results led us to calculate COV for anthropogenic deposits and the quaternary terrace separately.
F I G U R E 9 Cross-section of deposit models from the archaeological site of Santa Margarida (a, b) and the external surveyed area (c-e).
In NS cross-sections inside the archaeological area, anthropogenic deposits lay upon Miocene bedrock and Quaternary alluvial deposits. Deposit modelling in the non-excavated external area allowed us to distinguish sublayers within the anthropogenic deposit together with ancient soil surfaces [Color figure can be viewed at wileyonlinelibrary.com] The successful application of deposit modelling strategies (Bates & Bates, 2000, Carey et al., 2017, 2019  The interpretation of the data obtained has also been benefited from the determination of landform sediment assemblages (Layzell & Mandel, 2019). These are underlying genetically related packages of sediment and associated soils with predictable age relationships.
Bearing in mind that the archaeological record is a component of the sedimentary record, physical processes that remove, modify or bury sediments control the preservation and visibility of the record of the human past (Bettis et al., 1996, p. 2;Layzell et al., 2018, p. 1).
Identifying the stratigraphy of torrent flows and river terraces at this site in particular-there at least four river terraces known in the area of Martorell-distinguishing the Pleistocene bedrock, and knowing F I G U R E 10 Isoline map of the buried land-surface of the Quaternary deposits, showing the primal flow of the torrent in Santa Margarida before being drained the features of these alluvial landform assemblages informs about the potential areas in which earlier archaeological vestiges, if any, might have been preserved (Howard & Macklin, 1999). To that extent, geomorphological deposit modelling in Santa Margarida did not provide any evidence for the presence of archaeological vestiges within the floodplain sediment under the artificial ground. Certainly, given the limited exposure that a core provides, the physical analysis and description of core data are, in general, a poor way to identify buried archaeological materials/features. However, we know at this point that the site has almost no potential in its northern area since the garnet shale and Buntsandstein sandstone outcrops are closer to the surface.
Santa Margarida's long history has set a deep imprint on its surrounding landscape, understanding it as complex and wide-ranging entity that includes landform and environmental issues, but also the way in which human beings interact with them (Thomas, 2001). Santa Margarida's modern landscape-where a massive motorway broke the territorial unity of the Priory and its seventh-century-old buildings-is just the surface expression of these underlying processes of transformation (Price et al., 2011(Price et al., , p. 1059. Understanding the reality beneath required a wider geomorphological approach that fulfilled our initial aim of scheduling a more precise excavation plan, but it has also shed some light on the medieval and modern evolution of this site, which helped us understand how peasant agency and the landlords' interests interacted.

| CONCLUDING REMARKS AND FUTURE PROSPECTS
Geoarchaeological examination of Santa Margarida's subsurface has proved to be an effective way of defining the interface elevation between anthropogenic deposits and natural packages of alluvial sediment. This has opened brand new lines for data interpretation and has led to the discovery of ancient water flows, shedding new light on the primal layout of this site. In positively identifying this natural or geological horizon, we now have a broad idea about the total depth and volume of the excavation area before the fieldwork takes place, and we have obtained valuable information for interpreting post-depositional affectation of archaeological vestiges more precisely.
A combined approach of borehole logging, core sampling and dynamic probing on the basis of previous non-intrusive GPR testing has turned out to be a useful strategy for archaeological deposit modelling. A 10-year-long survey plan within a consolidated research project has allowed us to know the features of alluvial terraces and sediments therein, which enables us to decide to what extent the archaeological excavation is completed or not, and to design more cost-effective archaeological fieldwork schedules.
Further research in Santa Margarida is expected to complete the diachronic view of the landscape transformation in a rural area of Catalonia, from the late antique period to the present day. After a long research plan focussed on the site's layout and evolution, now it is time for a spatial turn attempting to understand the interaction between Santa Margarida's parish and the Monastery of Sant Genís, and the role of both as agents for construction of a feudal territory in Medieval Catalonia and its lordship endurance.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. They are part of unpublished technical reports (Tella, 2011(Tella, , 2019 kept in the Centre d'Estudis Martorellencs' archive.

PEER REVIEW
The peer review history for this article is available at https://publons.