Tracing the spatial organization and activity zones of an Early Mediaeval homestead at the Pohansko stronghold (Czechia) by combining geophysics and geochemical mapping

Geoarchaeological prospection techniques were applied to identify activity zones and the inner structure of a homestead at the Early Mediaeval site Pohansko near Břeclav (Czechia). By a combination of geophysical methods, the spatial distribution of microartefacts, geochemical analysis and multivariate statistical analysis, we outlined various manifestations of anthropogenic activity. We examined obtained data by Spearman's correlation coefficient, spatial autocorrelation (Global Moran's I) and robust Principal component analysis to identify the spatial pattern of the area. Recognized joint presence of heavy metals (Pb, Zn and Cu) and elements related mostly to organic matter, waste and ashes (S, P and Ca) as well as a small number of slag fragments probably indicate presence of metalworking zones or mixed zones with domestic and industrial debris at the homestead. Further anthropogenic activities could be connected to manuring, animal management or some kind of production activities based on the presence of Mn, P and Cu. Bone and charcoal concentrations supplement the information of geochemical analysis and may indicate the manner of waste management in the peripheral parts of the homestead. In the middle of the homestead, the location of archaeological features indicates an open space in which no specific activity was detected. By means of magnetic susceptibility and judging from the presence of daub, we defined the potential presence of non‐sunken features, which were not recognized by magnetometry. The outcome of the study is yet to be verified by excavation.

The main advantage of geophysical prospection is that extensive areas can be examined in a relatively short time, which facilitates the identification of a wide range of archaeological features (Aspinall et al., 2008;Gaffney, 2008;Milo, 2014).Geochemical surveying enables us to identify the organization of the space and distinguish boundaries or the particular functions of objects and structures, which significantly helps us to understand land-use activities on archaeological sites (e.g., Janovský & Horák, 2018;Salisbury, 2013;Save et al., 2020).The use of these methods can be especially important in the study of archaeological sites where the objects were situated at the level of the Early Mediaeval surface (and thus were not sunken into the subsoil) and their remains have not been preserved due to unfavourable environmental conditions or where the infilling of archaeological objects cannot be distinguished from the surrounding cultural layer (Seren et al., 2013;Simniškytė et al., 2021).The recognition of above-ground features and features not sunken into the subsoil has been a long-term subject of discussion in Early Mediaeval archaeology (Donat, 1980;Milo, 2014;Šalkovský, 2001).This issue is greatly relevant also at the Early Mediaeval site of Pohansko near Břeclav (Czechia), which represents one of the important centres of the Great Moravian Empire (Herold, 2012).Several archaeological features of various sizes and functions have been recognized in the area of the Pohansko stronghold excavated so far (Macháček, 2010).Their spatial distribution indicates a regular structure consisting of dozens of separate settlement units of rectangular shape, often referred to as farmsteads, households or homesteads (Dostál, 1982;Macháček, 2010).Such urban planning is unique in the context of Central European strongholds (Herold, 2012;Priš ťáková & Milo, 2021).The inner structure of homesteads consisted of several archaeological features located mostly along the edge of the homestead.In the middle, there was an open space the function of which is assumed to be communication or working, and so on (Macháček, 2010).Although the spatial organization of the stronghold has been explored and verified by a geophysical survey (Priš ťáková & Milo, 2021), the homesteads themselves and their internal organization have not yet been examined in detail.It has been suggested that these homesteads could have had a residential/ production function (Dostál, 1985;Macháček, 2010), but an analysis of agricultural tools has shown that agricultural production cannot be ruled out (Dresler & Beran, 2019).
Here, we provide the multivariate statistical and spatial assessment of the geochemical proxy record compared with geophysical results, including magnetic susceptibility signal, and the distribution of microartefacts at a selected homestead of the Early Mediaeval site of Pohansko.The main aim is to explore unrecognized spatial structures and identify different activity zones, thus gaining a closer insight into the function of the homestead and enriching our knowledge about life at the Pohansko stronghold.The means to achieve this are (1) the identification of archaeological structures and objects and the resulting definition of the internal spatial organization of the selected homestead and (2) the identification of zones of activities and their interpretation in terms of function.

| SITE DESCRIPTION
The archaeological site Pohansko is located in the southeastern part of Czechia, approximately 2 km southeast of the town of Břeclav.The site is located on the floodplain of the rivers of Dyje and Morava near the border of Slovakia and Austria (Figure 1).From a geomorphological point of view, Pohansko is part of the Lower Morava Valley, which belongs to the Vienna Basin (Demek & Balatka, 1987).The geology of the study area consists of Neogene deposits of the Vienna Basin, which are covered by Quaternary sediments mainly of fluvial origin (e.g., Havlíček et al., 2016;Nehyba et al., 2018).Important elements in the landscape are sandy elevations locally known as 'hrúdy', on one of which Pohansko is located.The elevation at the site ranges from 155 to 157 m above sea level.(Macháček, 2005;Macháček & Goláň, 2004).An Arenic Chernozem has been newly recognized in the area of the central part of Pohansko, with Gleyic Fluvisol in its immediate vicinity (Adameková, 2021).
The Pohansko stronghold has been archaeologically excavated for more than 60 years.There are traces of settlement from the Neolithic period to the 11th century.The greatest expansion of the settlement occurred between the second half of the 9th century and the beginning of the 10th century, when a fortified settlement, which spread over an area of about 50-60 ha, was built here (Macháček, 2005(Macháček, , 2010)).
The area of the site consists of a central fortified part and two fortified baileys.The central area is the size of 28 ha, out of which 4.66 ha has been archaeologically excavated, and is fortified with a 2-km-long wood and timber rampart with a stone front wall (Dresler, 2011).A total of 858 archaeological features and 12 homesteads have been identified by the most recent geophysical survey (Priš ťáková & Milo, 2021).Moreover, another seven square or slightly longitudinal homesteads of dimensions from 35 up to 55 m have already been excavated.A settlement pattern in the form of enclosed settlement units has few parallels, mostly in western or northern Early Mediaeval Europe, for example, at the sites Hamwic (Andrews, 1997), Haithabu (von Carnap-Bornheim et al., 2007) or Kirchheim (Geisler, 1997).The inner structure of the sites possibly reflects the organization of society and the function of fortified settlements or a controlled urbanization concept.In the Frankish Empire and neighbouring regions, it is connected to the merchants' and craftsmen's settlements with a combined residential/production function (Schulze, 1998).
The archaeological features recognized at homesteads consist of sunken-floored features of various functions, postholes, sunken-floored dwellings, hearths and in some cases even graves or preserved parts of palisade trenches.Some of the features were built on the original surface, possibly of timber construction (Macháček, 2010).Milo, 2021).The surveyed homestead is marked with a red line.specialized blacksmithing or non-ferrous metal production (Macháček, 2010), but excavated homesteads were generally connected to non-specialized home textile production (Březinová & Přichystalová, 2014), as well as carpentry, cooperage, pottery, and so on (Dostál, 1993).For most homesteads, agricultural production should be assumed.It is characterized by a complete range of agricultural tools in various stages of wear and preservation.They are found in all areas of the site (Dresler & Beran, 2019).Large osteological assemblages with evidence of stables and breeding are also common (Dreslerová, 2018).Based on that knowledge, we decided to noninvasively define the internal spatial organization of a particular homestead, identify zones of activities and interpret them in terms of function.

| Research design
In the first step, we selected one of the homesteads based on its overall properties and readability of magnetometric data obtained from previous research of the Pohansko stronghold (Priš ťáková & Milo, 2021).In the second step, we performed core prospection of a 46 m Â 42.5 m ($0.2 ha) area spanning the entire homestead (35 m Â 35 m) and its immediate vicinity (Figures 1 and 2) to obtain information about magnetic susceptibility, geochemical signature and microartefact distribution.
A total of 87 cores were described, documented and sampled in a 5 m Â 5 m square grid (Figures 2 and 3).Coring was performed using a Single Gouge Auger Set Eijkelkamp (diameter of 3 cm).Samples were collected according to the soil stratigraphy in the study area where Arenic Chernozem (having a texture class of sand in a layer ≥30 cm thick; IUSS Working Group WRB, 2014) was developed.The soil profile is characterized by a very dark brown A horizon, a very dark brown Cultural layer, a brown A/C horizon and a pale brown C horizon, with a sandy texture (Adameková, 2021).The maximum depth of the A horizon was 20 cm.The cultural layer was variable below the A horizon and ranged in depth from 13 cm to a maximum of 40 cm, with an average thickness of 24 cm.The depth of the A/C horizon ranged from 35 to 80 cm.In some cases, we hit an archaeological feature under the cultural layer, which possibly reaches a depth of more than 100 cm at the site (Macháček et al., 2021).
F I G U R E 2 Overview of core samples and their collection using an Edelman Auger Eijkelkamp in the area of a homestead at the Early Mediaeval stronghold Pohansko.The contour density is 0.1 m.
For the purpose of this study, we collected a total of 87 samples from the cultural layer, mostly from depths of 20-35 cm (i.e., the level of the Early Mediaeval surface) by coring for geochemical and magnetic susceptibility analyses.Moreover, we removed 0.5 L of material from the same level ($20-40 cm) and points using an Edelman Auger Eijkelkamp (diameter of 7 cm) to obtain archaeological material for microartefact analysis.Lastly, we used statistical methods to evaluate the obtained data.

| Geophysical survey
Based on long-term geophysical surveys of Early Mediaeval strongholds in Central and Eastern Europe, magnetometry has proven to be the most suitable geophysical method (Křivánek, 2018;Milo, 2014;Milo, 2019).A magnetometric survey of the Pohansko site was performed in the years 2011-2012 using a Förster Ferex 4.032 DLG magnetometer equipped with four probes.The method is based on the monitoring of local variations of the Earth's magnetic field.The inhomogeneities (magnetic anomalies) identified are caused by different ratios of ferromagnetic materials in monitored features and the enclosing soil matrix (Fassbinder, 2017).This method allows us to recognize features, including sunken featured dwellings, trenches and burnt features such as fireplaces, hearths or ovens (Fassbinder, 2015).
We used a density of 0.5 m between the individual profiles, which made it possible to detect smaller objects and to refine their interpretation.Based on the physical properties of anomalies, their shapes, dimensions and contexts, individual archaeological features and structures were mapped in a geographic information system (GIS) environment (Figure 4, left).

| Magnetic susceptibility
Magnetic susceptibility is a physical property defined as the ability of materials to magnetize in an induced magnetic field, which can be influenced by both natural and anthropogenic factors (Dalan, 2017;Mullins, 1977).It has been shown that this method can be successfully applied to archaeological sites to characterize differences in the magnetic properties of various archaeological features, or, as the case may be, their spatial distribution (e.g., Conger & Birch, 2019;Dalan, 2008;Křivánek, 2008).We applied mass-specific magnetic susceptibility (MS) to cored samples from the cultural layer (see Section 3.1) using an Agico Kappabridge MFK1-FA device set to operate with a magnetic field amplitude of 200 A/m.The samples were firstly air-dried, placed in diamagnetic plastic boxes and then measured at a low frequency of 976 Hz.Values of MS were expressed in m 3 Ákg À1 , projected in GIS (Figure 4, right) and interpolated using kriging (Oliver & Webster, 1990).

| Microartefact analysis
Microartefact analysis (MAA) is considered an effective tool for distinguishing residues of past human activities and determining the locations and possible nature of activity areas (e.g., Rosen, 1993;Sherwood et al., 1995;Stafford, 1995).MAA has been applied to  et al., 2018;Parker & Sharratt, 2017) and makes it possible to study smaller volumes of sediment (Dunnell & Stein, 1989;Stafford, 1995).
In some cases, MAA can offer a different and complementary dataset about archaeological sites (Homsey-Messer & Ortmann, 2016).Microartefacts are generally defined as man-made artefacts smaller than $1 cm.We obtained microartefacts from the cultural layer (see Section 3.1) using the water flotation technique on sieves with a mesh size of 1000 and 250 μm and manually selected.Fragments of ceramics, daub, bones, charcoal and slag were found.The quantified microartefacts were spatially plotted in GIS (Figure 5).

| Geochemical analysis
Samples for geochemical analysis were first dried and cleared of stones and organic residues.Then the samples were homogenized and powdered in an agate mortar.Thus prepared, they were pressed into pellets and finally analysed by an energy-dispersive fluorescence spectrometer (ED-XRF) Rigaku NexCG, which has a 50 W Pd tube and a silicon drift detector with a resolution of 145 eV.The device uses excitation of secondary targets to improve the signal-to-noise ratio.The time of excitation was 120 s for each target.Matrixmatching calibration of ED-XRF was done according to international reference materials and standards (e.g., GBW 07406, GBW 07103, SARM42, JSO1, DC 61101, GBW 03103, GBW 03101a, GBW 03102a, DC 78302, SRM2709a, BCR723, Metranal-31, Metranal-33, ERM-CC020, BAM-U110, NIST679 and SM9939) allowed quantification of 17 elements (Al, Si, P, S, K, Ca, Ti, Mn, Fe, Ni, Cu, Zn, Rb, Sr, Sb, Ba and Pb).All measurements are listed in Supporting Information S2.

| Data analysis and GIS interpolation
Input data, saved as CSV files, were analysed with R statistical software coupled with the RStudio environment (R Core Team, 2020).
The R libraries ggplot2 and factoextra were employed for data plotting and visualization.Summary characteristics were calculated using the R library psych.Spearman's correlation coefficient was calculated and visualized using the R libraries Hmisc and ggcorrplot.As a measure of spatial autocorrelation, global Moran's I index was calculated using the R library lctools.This index was used to evaluate whether the data were randomly dispersed.The robust Principal components analysis (PCA) for compositional data with the rrcov and robComposition libraries (Filzmoser et al., 2018) was applied to handle dataset with outliers.Element concentrations and principal components were projected in ArcGIS Pro 2.8.3 using the kriging function (Oliver & Webster, 1990).This interpolation method is considered highly suitable for visualizing measurements of soil properties (e.g., Salisbury, 2013;Wells, 2010).The search radius was set to eight neighbouring points, with the ordinary kriging method and the Out of 87 cores, nine cores should be placed directly in the geophysically recognized archaeological feature (Figure 2).The coring intersected sunken archaeological features 1/2, 5, 6, 12, 18 and 19 (Figure 3 and Table 1).Archaeological objects were manifested in the cores by exceeding the maximum depth of cultural layer and the distinctive boundaries with C horizon (Supporting Information S1).
The depth of the features ranged from 65 to 110 cm from the recent surface.The other archaeological features sunken in the subsoil, Nos 3, 10 and 16, were not confirmed.In the other three cores (GF48, GF83 and GF87), sunken archaeological features were captured but not found by geophysical methods.The depth of the features in the cores ranged from 53 to 65 cm from the present-day surface.
The most common was daub with 122 found pieces, followed by bones (48), charcoal (45), ceramic (5) and slag (3).Daub was found in 31 cores.The greatest amount of daub ( 18    Archaeological features interpreted based on magnetometry as well as microartefacts are located mostly on the margins of the homestead, which points to an open space in its middle (Figure 9).This space has been assumed to be an area for handling, walking, working or meeting Interpolations of the content of selected elements in the area of the prospected homestead at the Early Mediaeval stronghold Pohansko.Values in ppm.(Macháček, 2010;Macháček et al., 2021).This internal pattern, with its size and rectangular shape, is typical for the homesteads recognized at the Pohansko site.Despite this internal similarity, the built-up area and archaeological features are not completely identical in every homestead (Macháček, 2010;Macháček et al., 2021;Priš ťáková & Milo, 2021).
Coring within nine geophysically assumed archaeological features (Table 1 and Supporting Information S1) confirmed six sunken features (1/2, 5, 6, 12, 18 and 19) in the surveyed homestead.The cores in features 10 and 16 are both located at the very border of the potential archaeological feature, so they potentially missed the feature itself or they could represent features not sunken into the subsoil, the signal of which is too shallow but still visible in magnetometry (Seren et al., 2013).The recognition of non-sunken features at the Pohansko stronghold is complicated because their remnants are almost indistinguishable from the surrounding cultural layer.They are generally recognized based on the presence of stake holes, post holes, trenches, stone and mortar underpinnings, or ovens/fireplaces (e.g., Dostál, 1976;Vignatiová, 1992) as well as by analysing the spatial distribution of artefacts (e.g., Macháček et al., 2021).Feature 3, represented by a significant anomaly with high nT values, could potentially be a recent disruption or a hearth, oven or fireplace.The core sample, however, does not contain any traces of a sunken feature or charcoals, and values of MS are not significantly increased.It is known that geophysical anomalies, especially those caused by high temperature, may cover a larger area and often do not correspond to the exact shape of the object (e.g., Fassbinder, 2017;Tencer, 2019, 299).
Therefore, if the feature representing the anomaly is, in fact, minor and is not 4 m Â 5 m in size, it is possible that the core was drilled outside of the anomaly.Given the high nT values (over 20 nT), which could point to use of the fire (Fassbinder, 2015), together with archaeological features configuration known from previous archaeological excavations at the site (Tencer, 2019, 237, 331-332), we assume the presence of a hearth, oven or fireplace (Figure 9).Among the cores that were in the areas without geophysically recognized archaeological features, three different cores showed a presence of sunken features (Table 1).They have aboveaverage values of MS, even compared to most of the cores in the geophysically recognized features.None of them contained microartefacts.
All the mentioned cores are in line with other linear geophysically recognized features, two with feature 17 and one with feature 20.Therefore, we suppose they are connected directly to them (Figure 9).The anomaly caused by magnetic material present in the infilling of the archaeological object is not evenly distributed over the whole object (Tencer, 2019, 299-300), so unrecognized parts of the object captured in the cores did not provide values high enough to distinguish them from the surroundings and therefore did not appear on the magnetogram.
Other potential archaeological features may be indicated by increased values of MS in combination with the spatial distribution of microartefacts.High values of MS in archaeological contexts are particularly connected to burned materials like daub, slag, charcoal, and so on, features destroyed by fire or places of production.Fillings of archaeological features can also provide values sufficiently contrasting with the surrounding cultural layer (Křivánek, 2008;Milo, 2013).In the northern corner of the homestead, between features 1, 13, 15 and 16, values of MS increase significantly.Together with numerous fragments of charcoal and fewer ones of daub, we suggest the presence of a non-sunken or indistinguishable feature in the area.
Another area with increased values is in the western corner between features 11, 20 and 12, and partially in the eastern corner of the homestead (Figure 9).Increased values in the western corner could possibly be connected to the small geophysically recognized feature 11 (Figure 4).The potential presence of an archaeological feature is supported by a strong accumulation of daub in both areas, which could be linked to the destruction of an above-ground building (Macháček, 2010).The actual shape, orientation or dimensions of the above-ground features cannot be determined from the survey data and are therefore based on information obtained from the archaeologically excavated areas (Macháček et al., 2021, 35-41).
F I G U R E 8 Interpolation map of the first, second and third principal components of the prospected homestead at the Early Mediaeval stronghold Pohansko.

| Spatial patterns and interpretations
The approach of non-invasive geochemical surveying has already been applied with the purpose of exploring ancient activities at archaeological sites in Eastern-Central Europe.It has been demonstrated that it is a cost-effective tool that can help reveal anthropogenic geochemical signal of settlement activities or agricultural practices of different ages (e.g., Danielisová et al., 2022;Horák et al., 2018;Horák & Klír, 2017;Janovský et al., 2020;Janovský & Horák, 2018;Salisbury, 2013;Šantru ˚čková et al., 2020).The design of research varies in these types of surveys (e.g., the depth sample collection, the soil horizons sampled, the grid of the sample area) and are adapted to local caused by anthropogenic sources that will be discussed below.PC2 is, according to the detected elements (Rb, K, Ti, Mn and Al), possibly of a geogenic nature (e.g., Horák & Klír, 2017), and therefore, we will not interpret it in more detail.PC3 indicates a negative correlation between hydromorphic/pedogenic feature (Mn with positive values) corresponding with digital elevation model (Figure 2) and some elements (S, Ba, Sb and P) associated with human activities concentrated in specific features (1, 2, 3, 4, 5 and 6).
PC1 is linked to heavy metals, such as Pb, Zn and Cu, which may in archaeological contexts indicate some kind of metalworking activity (Aston et al., 1998;Cook et al., 2010;Danielisová et al., 2022) and to Ca, S and P relating mostly to organic matter, body waste, ashes and bones (e.g., Cook et al., 2014;Eriksen et al., 1998;Holliday & Gartner, 2007;Proudfoot, 1976;Schlezinger & Howes, 2000).In general, the joint presence of these elements probably reflects the bonding of heavy metals with organic matter (Kwiatkovska-Malina, 2018).
This association relates to archaeological features 12 and 20 and at the same time some of the related elements (P and Zn) having significant Moran's I index values indicating specific spatial distribution.
Altogether it could indicate presence of metalworking zones or mixed zones with domestic and industrial debris (Dirix et al., 2013).The strong concentration is also recorded in the northeastern part of the surveyed area, between features 2 and 18.The small number of slags found through microartefact analysis may also support the idea of metalworking activity at the homestead.Based on archaeological excavations, we can conclude that the slag was located in the wider vicinity of the buildings where metalworking activities have been documented.However, its scatter reaches up to 20 m from the production objects (Priš ťáková, 2022, 184-185).Preliminarily, we can cautiously assume it indicates non-ferrous metal processing in certain parts of the homestead, probably on a small scale because the values of the individual elements were not high.The processing of nonferrous metals is documented in a few homesteads on the site.It is recorded by the presence of fireplaces and heating devices together with artefacts related to non-ferrous metal production, for example, crucibles, small anvils and pieces of slag (Macháček et al., 2007).
The third principal component (PC3) illustrates the negative correlation between the elements of Ca, Fe, Al and Sr against Mn, P, Sb and Cu.Considering the negative correlation of Mn with P and Sb on the one hand and Ca, Fe and Al on the other hand, we can assume that PC3 could be related to different redox conditions in sediments (negative correlation of Fe with Mn).At the same time it could be related to anthropogenic influence (P and Ca), both options can be related to specific conditions in buried archaeological objects.In the case of anthropogenic influence, the presence of Mn, P and Cu may point to the animal waste but also to ashes and possible metallurgy (Bintliff & Degryse, 2022;Nielsen & Kristiansen, 2014).Thus, the high negative values of PC3 could be linked to the archaeological settlement structure, which may not only be related to manuring or animal management but also to production activities.The highest values are in the vicinity of the features 12 and 20, which are especially interesting as there is a concentration of positive PC1 values around them.A smaller concentration is also found in the vicinity of a strong geophysical anomaly (feature 3) in the eastern corner of the homestead.If the anomaly is a remnant of a fireplace (see Section 5.1), the increased amount of phosphorus could be interpreted as an accumulation of food preparation waste or the use of dry dung as fuel.Bone concentrations are not associated with any of the components, which is rather surprising, as their location on the edges and corners of the homestead would suggest their intentional deposition and point to waste management, which could correspond with the elevation of the selected features.We observe the same situation in the case of charcoal distribution (Figure 5).However, this can be significantly influ- space), to reveal and compare the components outside of them (e.g., communications) and to prepare basic knowledge for a precisely targeted archaeological field survey.
The spatial distribution of waste and the nature of the moveable and non-moveable finds from excavated parts of the Pohansko site indicate different zones connected with activities at the homestead (Priš ťáková, 2022).Some homesteads were associated with F I G U R E 1 Location of the Early Mediaeval site Pohansko near the town of Břeclav (Czechia), the central district and the north-east suburb of the stronghold.Interpretation model of the stronghold's urban planning according to results of magnetometry and previous archaeological excavations (modified after Priš ťáková &

F
I G U R E 3 Comparison of documented cores with recognized soil horizons and the cultural layer.The full composite photography of 87 documented cores from the surveyed area of the homestead at the Early Mediaeval stronghold Pohansko can be found in Supporting Information S1. many archaeological sites (Homsey-Messer & Humkey, 2016; Parker

F
I G U R E 4 Results of magnetometric survey with numbers of the features (left) and interpolation map of magnetic susceptibility (right) of a selected homestead at the Early Mediaeval stronghold Pohansko.spherical semivariogram model.The search radius was chosen to follow the relationships of surrounding points, but at the same time, it was not too generalized.4 | RESULTS 4.1 | Magnetometry and core prospection A total of 20 potential archaeological features were identified in the study area, 16 of which are located within the expected area of the homestead, mostly on the sides of the homestead.The potential archaeological features are characterized by slightly positive magnetic values (2-10 nT) compared to the surrounding values (À0.2/+0.5 nT).This observation has been verified by previous archaeological excavations.In total, 14 such positive structures (1-2, 4-13, 15-16) have been recorded inside the proposed homestead area and another 4 (17-20) outside of the homestead within the coring area (Figure4).One potential archaeological feature (14) is slightly negative with À2 nT, located close to the northwestern border of the homestead.An anomaly, located in the eastern corner of the homestead (3), shows high magnetic values (À50/+130 nT).Visible bipolar anomalies (À50/+50 nT) located mostly outside the homestead have already been interpreted as magnetic metal objects connected to recent disruptions(Priš ťáková & Milo, 2021).Therefore, their interpretation is not included in the layout.
interpolated result of MS shows considerably high values (621-338 m 3 Ákg À1 ) in the northern corner of the reconstructed homestead, between the cluster of smaller archaeological features 13-16 and trench-like features 1/2.Other higher values (337-250 m 3 Ákg À1 ) are F I G U R E 5 Interpolation map of microartefact spatial distribution (left) and distribution of individual categories of microartefacts (right) in the area of the study homestead at the Early Mediaeval stronghold Pohansko.located in the northern and western parts and on the eastern border of the surveyed area, again close and between possible archaeological features 11, 12, 17 and 20.Values of 249-172 m 3 Ákg À1 cover solid areas in the south-western, western, northern and eastern parts.Most of them are connected to the empty spaces but also to potential archaeological features 3, 4, 10, 12 and 18.The lowest values (<172 m 3 Ákg À1 ) are located in the central part and continue to the south-western, southern and partly south-eastern parts of the area ) was found in the western corner of the homestead near feature 11.An increased amount of daub was also found around nearby features 10 and 12. Another concentration of daub is located at the eastern border of the surveyed area.A smaller concentration in the north-eastern part, between archaeological features 1 and 17, is also noticeable.The northern, northwestern and southeastern parts are clear of any microartefacts.Bones were found in 22 cores.Most of them come from the southern half of the surveyed area.Their highest concentration is around the southern edge of the homestead, around features 7-9 and it continues around the southeastern border towards features 6 and 3. Another concentration of bones is around the western corner around features 11 and 20.In the northern part, bones were found only between features 1 and 17.In the central and northwestern parts, no bones were found.Charcoal was found in 14 cores.Its spatial distribution was mostly connected to the northern half of the surveyed area, in the vicinity of features 1, 2, 12, 16 and 18.Two finds are located in the middle of the homestead.A slightly greater amount of charcoal was found above feature 6.In the western part, there are three finds near feature 20 and in the southern one find close to features 7 and 8. Ceramic finds are surprisingly scarce.They were found only in four cores; two pieces come from above feature 5, one comes from the western part, close to feature 20, and the last two cores are in the north.Small pieces of slag were found in three cores, one in the south, near feature 8, one in the central part, and one in the east, outside of the reconstructed homestead.The overall spatial distribution shows a strong connection of the microartefacts to the presence of the archaeological features and to their vicinity (Figure5).Especially high values are above features 6, 10 and 11.Noticeable accumulations are also around features 1, 2, 9, 12, 16, 17 and 20, around the borders of the reconstructed homestead.Empty spaces are at the northwestern and northeastern borders of the surveyed area, slightly north from the central part and in the south.4.4 | Geochemical data and multivariate statistics Values of 17 chemical elements, MS and number of microartefacts were used for calculation of Spearman's rank correlation coefficient, spatial autocorrelation global Moran's I and summary statistics (Supporting Information S2).Correlation coefficients show relatively weak correlations between individual elements (Figure 6).A high number of weak correlations usually results in a higher number of main components representing a given amount of total variance of the data.The T A B L E 1 most significant positive correlations are between the elements of Pb and S, Cu and Zn, Ba and Sb and between the elements of Si and Al (Figure 6 and Supporting Information S2).P positively correlates with Ca, Cu, Zn, Pb and S, which significantly negatively correlates with K and positively with Ca and P. Ba is positively correlated with Sb, and Sr and Zn are positively correlated with S, Cu and Pb.The element of Mn is strongly positively correlated with values of MS, which positively correlates with P and negatively with Fe.Regarding microartefacts daub positively correlates with Pb, S, P and Ni, charcoals positively correlate with MS and Mn and negatively with Fe and bones negatively correlate with Mn a MS.Elements with Moran's I index close to zero, specifically Al and Sb, show a random spatial distribution within the area studied.There is also a visible tendency to form local accumulations of elements with positive Moran's I index, especially in the case of Mn, Zn, P and MS while there are negative values of charcoals and low values of Sb, Al and daub (Supporting Information S2).The results of chemical elements are also visualized in the space of the homestead (Figure 7; see Supporting Information S2).The spatial distribution of the individual elements shows great variability.Elements like Si, K, Ti and Ni tend to create a significant number of scattered accumulations, while Cu, Sr, Ba or S cover more coherent areas.P, Ca and Mn create only a small number of spots with higher values.Al, P, K, Ca, Mn, Fe, Cu, Rb and Zn are mostly connected to areas near the borders of the homestead.In some of the geophysically recognized archaeological features, there are concentrations of P, Ca, Mn, Fe, Ni, Cu, As, Pb and Zn.In the vicinity of the features are also higher values of Si, S, Ti, Rb, Sr and Ba.A very different trend is visible with the elements Sr, Ba and partly S, which are connected mostly to the central part of the homestead without any recognized archaeological features.Out of all the elements analysed, only P shows a very strong tendency for low values in the central part of the homestead.A similar tendency is also partly visible in the spatial distribution of Ca and Zn.The results of the analysis of geochemical concentration were processed by robust compositional PCA, which includes 17 variables for 87 samples.The first three main components explaining $75% of the total variance in the dataset were selected according to the scree plot (see Supporting Information S2) and are analysed here.The first principal component (PC1) with 36.9% of the variance is influenced mostly by elements of S, P, Ca, Zn, Cu, Pb which negatively correlated with Ba, Sb, Sr, Al and Ni (Supporting Information S2).Positive values are dominantly located along the eastern, southwestern and northwestern borders and partly at southern and northern borders, mainly beyond the presumed boundaries of the homestead (Figure9).The areas with the greatest concentrations are around features 12 and 20, between features 2, 3 and 18 and between features 10 and 19 (Figure8).Negative values create a significant concentration in the centre of the homestead, around feature 1 and 16 and partly beyond the northern border of the homestead.The second principal component (PC2) with 24.6% of the variance is linked with Mn, Al, K, Ti, Rb against S, Ba, Sb, P (Supporting Information S2).Positive values are represented mainly beyond the presumed boundaries of the homestead in the northeastern and eastern part of the surveyed area and slightly in the north between features 12, 14, 15 and 16.Negative values are again significantly concentrated in the central part of the F I G U R E 6 Matrix plot of Spearman's rank correlation coefficient with grouped variables.The size and colour intensity of the ellipses indicate the absolute value of the correlation coefficients.For exact values, see Supporting Information S2. homestead, around features 1, 2, 3, 4, 5 and 6.Another concentration of negative values is located around features 11, 19 and 20 in the eastern part of the surveyed area (Figure 8).The third principal component (PC3), with 13.8% of variance, illustrates the relation between the elements Ca, Fe, Al, Sr against Mn, P, Sb, Cu (Supporting Information S2).Positive values are significantly located in the southern part of the surveyed area and in the vicinity of features 1 and 2. Negative values are located in the northern part of the surveyed area and around features 3, 12, 18 and 20 (Figure 8).

|
Interpretation of geophysical and geoarchaeological survey

F
I G U R E 9 Resulting interpretation model with recognized zones of activity and features at the homestead at the Early Mediaeval stronghold Pohansko.The model is based on the results of spatial distribution of geochemical data and multivariate statistics, microartefacts, magnetometry and MS.environmental conditions, the geological background and to the research questions asked.A universal methodology does not exist, because there are numerous diverse factors or disturbances within different archaeological sites or time periods, so it is necessary to approach each site individually.Multivariate spatial analysis of the cultural layer revealed different zones associated with different activities and functions.The areas of high positive values of PC1 (S, P, Ca, Zn, Cu and Pb) are potentially enced by the chosen grid of sampling and the volume of samples of soil for the flotation (see Section 3.1).Bone and charcoal concentrations supplement the information from geochemical analysis and may indicate how the waste is managed at the homestead.They are mainly concentrated in the peripheral parts of the homestead.This correlates with the concentrations of chemical elements that are also found mainly in the peripheral parts and around the objects.All this information together allowed us to reveal an open space in the middle of the homestead (Figure9).6| CONCLUSIONOur research demonstrates that certain limitations of geochemical mapping can be compensated by synergy of methods such as geophysical survey, magnetic susceptibility analysis and, last but not least, microartefact analysis.The new data on the spatial organization within the studied homestead brings insight into the non-excavated part of the Pohansko stronghold.Our results provide a basis for defining possible metalworking or mixed zones with domestic and industrial debris and areas with traces of manuring or animal management and production activities.Besides these activity areas, a fireplace or hearth/oven as well as possible presence of above-ground features was recognized from the collected data.In the middle of the homestead, the open space without sings of specific activity was determined.Regarding recognized inner structure of the homestead, we have identified several analogies with previously excavated homesteads.All the above-mentioned ideas will be tested in light of further archaeological excavation.Nevertheless, it is already apparent that the research presented here has provided information that would likely have been missed by a purely excavational study, specifically the identification of possible above-ground features from the surrounding cultural layer.Further geochemical sampling and analysis done on a smaller grid and over a larger area should follow to help understand the inner structure of the homesteads better (e.g., open PRI ŠŤ ÁKOV Á ET AL.
Table of the cores located within geophysically recognized features or with recognized features sunken into the subsoil.