Phoenician Pottery in the Western Mediterranean: A New Perspective Based on the Early Iron Age (800–550 BC) Settlement of Sant Jaume (Alcanar, Catalonia)

: One of the most important reception sites for Phoenician pottery imports in the NE Iberian Peninsula is the Early Iron Age (800–550 BC) settlement of Sant Jaume. This site is exceptional in terms of preservation and the large number of complete vessels recovered. Moreover, the ceramic assemblage comprises one of the best collections of the earliest wheel-thrown pottery that is considered evidence of trade from the western Phoenician colonies and their speciﬁc interest in exploiting metallurgical resources. In this research, a sample of 58 individuals of wheel-thrown pottery has been analysed by X-ray ﬂuorescence (XRF), X-ray diffraction (XRD), petrography (PE), and scanning electron microscopy attached with an energy dispersive X-ray unit (SEM-EDX). It was possible to identify 29 ceramic groups, some of which correspond to known Phoenician workshops of southern Andalusia and Ibiza, though the origin of most groups remains to be determined. The wide variety of sources identiﬁed illuminates the patterns of trade and exchange that the Phoenicians developed during the Early Iron Age and the export of their manufactured products. This information is fundamental to our understanding of the economic system developed by the Western Mediterranean Phoenician colonies that affected and transformed indigenous communities in the Mediterranean region.


Introduction
The Phoenician arrival in the Iberian Peninsula, attracted by the wealth of metal resources [1], has been the subject of intense debate, primarily because written and archaeological sources differ significantly.Although classical sources point to the foundation of Gadir (Cádiz) by the Tyrians in the 12th century BC, there is no archaeological evidence for the first Phoenician contacts with the southern Iberian Peninsula before the 10th century BC [2,3].Moreover, the Morro de Mezquitilla [4], La Rabanadilla [5], and the evidence from the Cine Cómico in Cádiz [6] cannot be used to date permanent settlements before the end of the 9th century BC or the beginning of the 8th century BC.At that time, the places used as stopovers on long-distance journeys in the Western Mediterranean became proper urban colonies [7,8] (Figure 1a,b).However, around the middle of the 7th century BC, there was a further expansion of these settlements into surrounding areas.The results were larger settlements, such as the well-documented examples in the region of Vélez-Málaga [9].
were larger settlements, such as the well-documented examples in the region of Vélez-Málaga [9].The archaeological campaigns undertaken between 1964 and 1984 by the Deutsches Archäologisches Institut (DAI) at the sites of Toscanos, Morro de Mezquitilla, Alarcón, Jardín, Trayamar, and Chorreras [10][11][12][13][14][15] offered the first 14 C dates related to stratigraphic contexts of the beginning of the 8th century BC for some of these early colonies.These excavations also enabled the recovery of large ceramic assemblages of Western Phoenician The archaeological campaigns undertaken between 1964 and 1984 by the Deutsches Archäologisches Institut (DAI) at the sites of Toscanos, Morro de Mezquitilla, Alarcón, Jardín, Trayamar, and Chorreras [10][11][12][13][14][15] offered the first 14 C dates related to stratigraphic contexts of the beginning of the 8th century BC for some of these early colonies.These excavations also enabled the recovery of large ceramic assemblages of Western Phoenician products, leading to a profound impact on the interpretation of the Early Iron Age archaeological record in the northeastern Iberian Peninsula.Traditionally, the development of the Iberian culture had been discussed in the light of the Greek colonisation, but the recognition Appl.Sci.2023, 13, 3733 3 of 29 of the intense activities of Phoenicians as pioneer traders acted as a timely counterbalance.In fact, the Phoenician presence in the northeastern Iberian Peninsula (Figure 1c), its associated trade network, and the interaction with indigenous communities did not become a key topic in the study of protohistory until the end of the 1960s [16].From then on, wheel-thrown pottery, identified by typological and macroscopic observations [17][18][19][20], was used as an indicator of interaction with the known Phoenician colonies.
To suggest the mechanisms that were prominent in the Phoenician economic system in the Western Mediterranean, it is essential to understand the dynamics between the southern colonies and the northeastern indigenous communities.It seems that the Phoenicians favoured long-distance and regular commerce, and we can identify this fluid and intense presence thanks to the recovery of the first wheel-thrown pottery in indigenous contexts from the 8th century BC to ca. 575 BC.However, we lack explicit knowledge of how this system operated, and there are still many questions to be answered about the early presence of Phoenician traders in this part of the Iberian Peninsula, especially whether they established themselves in the area in permanent settlements.Additionally, we must not forget the possible role played by the southeastern indigenous potters after the arrival of Phoenician pottery since, in some cases, wheel-thrown pottery was already being made in indigenous settlements by the end of the 8th century BC [21].The existence of such indigenous wheel-thrown pottery brings an added complexity that is only now beginning to be explored.
In 2002, the Universitat de Barcelona launched an analytical programme on Phoenician pottery recovered from different sites in Catalonia to shed light on its provenance [22].Through this work, Sant Jaume stood out as a critical settlement in understanding the variety of Phoenician products traded during the Early Iron Age, thanks to its extraordinary state of preservation and rich assemblage, which may provide some indications about the broader dynamics at work.The results of the Central Mediterranean Phoenician pottery imports have been discussed elsewhere [23], and they suggest that the network for distributing Phoenician goods was more complex than previously thought.
This paper analyses the Western Mediterranean Phoenician imports found at this site.One of the main objectives is to determine their provenance and examine whether they were manufactured in different areas or have a much more restricted source.Typological information is also considered in studying correlations between archaeometric and archaeological data.

Materials and Methods
Sant Jaume (Alcanar, Montsià, Catalonia, Spain) (N 40 • 35 4.0596; E 0 • 28 34.1256) is an Early Iron Age settlement located on the top of a small hill (224 m AMSL) at the southern edge of the Montsià Mountains, 5 km from the mouth of the River Sénia, 20 km from the mouth of the River Ebro, and only 2 km from the coastline.It is a small site (around 700 m 2 ) with a remarkable defensive system [24] and a pseudo-circular plan with a single phase of occupation (6th-8th century BC).Our main hypothesis is that Sant Jaume was a fortified residence, perhaps the home and the headquarters of local political power in the framework of a chiefdom, from which it was possible to control neighbouring settlements (Moleta del Remei, la Ferradura, la Cogula, and el Castell d'Ulldecona).This chiefdom (Figure 1c) has been named the Sant Jaume Complex [25][26][27][28][29].However, the work carried out in recent years at the site suggests the possibility that it could be a Phoenician factory, though further research is needed to support this hypothesis.
None of the excavated remains on the site (Figure 2) can be considered domestic units.Instead, the two-storey buildings consisted of a lower level related to stabling, the reception and processing of agricultural and livestock products and other activities, while the upper floors were used as warehouses, storing large amounts of manufactured products, raw materials, and different types of objects, including a plethora of pottery.The ceramic assemblage at Sant Jaume can be divided into two main categories: hand-made pottery, similar to that of the Late Bronze Age [30], and wheel-thrown pottery, related to that of Phoenician colonies in the Western and Central Mediterranean Sea [23].The wheel-thrown assemblage is composed of: T.10.1.2.1 type amphorae [31] (Figure 3a2), a new amphora type labelled as T.2.1.1.2or similar (Figure 3a1), pithoi and pithoid vessels (Figure 3b), narrownecked cylindrical jars of the Cruz del Negro type (Figure 3c), and, in lesser quantities, tripod mortars/vessels (Figure 3d).Besides these common types that also appear in other Iron Age settlements identified in this area, Sant Jaume also has fine tableware: carinated bowls (Figure 3e,f), plates (Figure 3g,h), red slipped jugs (Figure 3i), dipper jugs (Figure 3j), and oil bottles (this last type, only recently recovered in the excavations, is still under study and is not yet represented in the general typological table of Phoenician pieces from the site in Figure 3).
Appl.Sci.2023, 13, x FOR PEER REVIEW 4 of 30 raw materials, and different types of objects, including a plethora of pottery.The ceramic assemblage at Sant Jaume can be divided into two main categories: hand-made pottery, similar to that of the Late Bronze Age [30], and wheel-thrown pottery, related to that of Phoenician colonies in the Western and Central Mediterranean Sea [23].The wheelthrown assemblage is composed of: T.10.1.2.1 type amphorae [31] (Figure 3a2), a new amphora type labelled as T.2.1.1.2or similar (Figure 3a1), pithoi and pithoid vessels (Figure 3b), narrow-necked cylindrical jars of the Cruz del Negro type (Figure 3c), and, in lesser quantities, tripod mortars/vessels (Figure 3d).Besides these common types that also appear in other Iron Age settlements identified in this area, Sant Jaume also has fine tableware: carinated bowls (Figure 3e,f), plates (Figure 3g,h), red slipped jugs (Figure 3i), dipper jugs (Figure 3j), and oil bottles (this last type, only recently recovered in the excavations, is still under study and is not yet represented in the general typological table of Phoenician pieces from the site in Figure 3).The amphorae are the most common type in the collection.They mostly match Type T.10.1.2.1, which is the most common type of ancient Phoenician work in the Western Mediterranean colonies.Nevertheless, other amphora types, clearly different from the latter "canonical" type, are represented by several individuals with a much more prominent and everted rim and a tendency to a cylindrical body.We also find notable differences and variations in the pithoi, related to their shape and size.The highest morphological diversity is present in the narrow-necked cylindrical jars, or Cruz del Negro type, with almost every individual exhibiting different characteristics.Finally, the tripod mortars/vessels also display a marked lack of standardisation in their production.
For our study, 58 samples were taken, including especially amphorae (32), pithoi (5), tripod mortars (6), jars (10), plates (2), dipper jugs (1), and undetermined ceramics (2) (Table 1).These analyses complement those already published for Central Mediterranean Phoenician imports [23] (4 individuals) and hand-made pottery [30] (15 individuals, though one hand-made vessel, recently characterised, remains unpublished).The amphorae are the most common type in the collection.They mostly match Type T.10.1.2.1, which is the most common type of ancient Phoenician work in the Western Mediterranean colonies.Nevertheless, other amphora types, clearly different from the latter "canonical" type, are represented by several individuals with a much more prominent and everted rim and a tendency to a cylindrical body.We also find notable differences and variations in the pithoi, related to their shape and size.The highest morphological diversity is present in the narrow-necked cylindrical jars, or Cruz del Negro type, with almost every individual exhibiting different characteristics.Finally, the tripod mortars/vessels also display a marked lack of standardisation in their production.
For our study, 58 samples were taken, including especially amphorae (32), pithoi (5), tripod mortars (6), jars (10), plates (2), dipper jugs (1), and undetermined ceramics (2) (Table 1).These analyses complement those already published for Central Mediterranean Phoenician imports [23] (4 individuals) and hand-made pottery [30] (15 individuals, though one hand-made vessel, recently characterised, remains unpublished).All individuals have been characterised chemically through wavelength dispersive Xray fluorescence (WD-XRF) and mineralogically through powder X-ray diffraction (PXRD).For analysis by WD-XRF, samples of about 15 g were taken from each individual.The superficial layers were mechanically removed, and the samples were milled in a tungsten carbide cell mill, the Spex Mixermodel 8000.The chemical composition was determined from powder dried in an oven for 12 h at 105 • C. Two 30 mm glass bead replicates were made by mixing 0.3 g of the dried sample with 5.7 g of lithium tetraborate (Li 2 B 4 O 7 ) flux (1/20 dilution) and 5 mg of lithium iodide (LiI) as a release agent to determine the major and minor elements.This mixture was homogenised, placed in a 95% Pt, 5% Au crucible, and melted in a fully automatic bead preparation system, PANalytical Perl'X-3, at 1125 • C. Pressed powder pellets were made using 6 g of specimen mixed with 2 mL of a binding agent solution of n-butyl methacrylate synthetic resin (Elvacite ® 2044) in acetone at 20% by mass to determine trace elements.This mixture was manually homogenised in an agate mortar to dryness and placed on a base of boric acid (H 3 BO 3 ) in an aluminium vessel of 40 mm diameter subjected to a pressure of 200 kN for a period of 60 s using a Herzog press.Due to the long duration of the research project, two different XRF instruments were used to determine the concentrations.First, in the case of individuals MOS001-MOS054, a Phillips PW2400 spectrometer with an Rh excitation source was used.Then, individuals MOS061-MOS083 were analysed using an Axios mAX -Advanced PANalytical spectrometer with an Rh excitation source.Both spectrometers were calibrated by a suite of 56 international Geological Standards.Interferences were considered, and matrix effects were corrected.For individuals MOS061 to MOS083, these corrections were made using the PANanalytical Pro-Trace software for trace elements.The elements determined were: Na 2 O, MgO, Al 2 O 3 , SiO 2 , P 2 O 5 , K 2 O, CaO, TiO 2 , V, Cr, MnO, Fe 2 O 3 (as total Fe), Co, Ni, Cu, Zn, Ga, Rb, Sr, Y, Zr, Nb, Mo, Sn, Ba, Ce, W, Pb, and Th.For individuals MOS001 to MOS054, Na 2 O was determined from the pressed powder pellet, while for individuals MOS061 to MOS083, it was determined from the glass bead replicates.Finally, for individuals MOS061 to MOS083, the trace elements Sc, La, Sm, and Yb were also determined using only PANalytical synthetic standards, and their concentrations must be considered semiquantitative.Major and minor elements are expressed as concentrations of oxides in mass fraction percentage (w%).Trace elements are expressed as elements in w mg/kg.Loss on ignition (LOI) (expressed as w%) was determined by firing 0.3 g of the dried specimen at 950 • C for 3 h.Calcinations were carried out in a Heraeus muffle model M-110 using a heating rate of 3.4 • C/min and free cooling.For the statistical data treatment, the concentrations of Mo and Sn were discarded due to analytical imprecision, as were those of Co and W because of possible contamination from the tungsten carbide cell of the mill used.P 2 O 5 concentrations were not used in the data treatment as some values were considered erratic, and some individuals were pushed out of their group.Such values can be due to contamination during burial, for example, from organic matter [32] (and references therein).However, the general influence of this element was low, and the main structure defined in the data set was preserved after it was discarded.Sc, La, Sm, and Yb were discarded for the general data treatment and only used for comparison with other projects.
The mineralogical characterisation of all individuals was performed using PXRD.The previously prepared powder specimens were manually side-loaded and pressed with frosted glass in a cylindrical sample holder of 27 mm diameter and 2.5 mm high (PW 1811/27).Measurements for individuals MOS001-MOS054 were made using a Bragg-Brentano geometry diffractometer, Siemens D-500, working with the Ni-filtered Cu-Kα radiation (λ = 1.5406Å) at a working power of 1.2 kW (45 kV and 30 mA), from (4 to 70) • 2θ, at 1 • 2θ/min (step size = 0.05 • 2θ; time = 3 s).Measurements for individuals MOS061-MOS083 were made using a Bragg-Brentano geometry diffractometer PANalytical X'Pert PRO MPD Alpha-1 (radius = 240 mm) equipped with an X'Celerator detector (active length = 2.122 • ), using the Ni-filtered Cu-Kα radiation (λ = 1.5406Å) at a working power of 45 kV and 40 mA, from (5 to 80) • 2θ, with a 0.026 • step size and an acquisition time of 50 s, spinning the sample at 1 Hz.The crystalline phases in each analysed specimen were evaluated using the PANalytical X Pert HighScore Plus software package, which includes the Powder Diffraction FileTM (PDF ® ) of the International Centre for Diffraction Data (ICDD).
Finally, a subsample of eleven individuals was studied by scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDX) (MOS-005, 007, 012, 023, 032, 048, 054, 061, 062, 068 and 069) according to the classification by WD-XRF and PXRD analysis of meaningful compositional groups, by petrographic criteria, and according to specimen availability [35].SEM-EDX observations were performed on a fresh cross-section fracture passing through the oro-aboral axis of the body wall to observe the microstructure, estimate the degree of sintering and the vitrification state of the matrix, and enable microanalysis of features of interest.Bulk specimens were fixed on metal specimen stubs using silicone adhesive, and the non-conductive ceramic specimens were made conductive.Colloidal silver paint was applied on the ceramic bulk specimen's excess silicone adhesive and lateral sides.Subsequently, the specimen surface was coated with a thin carbon film (~10 nm) by vacuum evaporation.The observations were made using JEOL JSM-6510 and Quanta 200 microscopes coupled with an energy-dispersive X-ray spectroscopy system (EDX).The observations were performed using an acceleration voltage of 20 kV and a working distance of 15 mm.

Results
The results of elemental concentrations determined by WD-XRF (Table 2) correspond with a particular case of the projective d + 1-dimensional space where the projective points are projected into the simplex S d .Points are represented by homogeneous coordinates that have a constant sum k (k ∈ R+), (in this current case, k = 100).The projective points' vector space is the positive orthant.Hence, for the statistical data treatment, the raw concentrations have been alr (additive log-ratio) transformed according to being S d the d-dimensional simplex, and They have also been clr (centred log-ratio) transformed following the equation being S d the d-dimensional simplex, g(x) the geometric mean of all d + 1 components of x and H ⊂ R d + 1 a hyperplane vector subspace of R d + 1 [32,[36][37][38][39].The statistical treatment of the chemical data was performed using R [40].The first analysis carried out was to calculate the variation matrix that completely determines the covariance structure of compositional data and provides the total variation (tv) of the analysed ceramic assemblage (Figure 4) [36,41].The overall data set exhibits a total variation of 2.28, a high value that indicates a polygenic group's existence.If the total variation quantifies how different the chemical data are, the information entropy, or Shannon index, quantifies how evenly the chemical differences are related to the retained components.Here, the chemical differences are related to many components (H 2 = 3.87 Sh; H 2 % = 84.41).As can be seen in the compositional evenness graph, the elements that introduce more variability are MnO and CaO (tv/τ .j< 0.3) and, to a lesser extent, Na 2 O, Cu, K 2 O, Rb, and Zn (0.3 < tv/τ .j< 0.5), but no components introduce low variability (tv/τ .j> 0.9).Thus, total variation and compositional evenness indicate a complex structure in the data set.
Appl.Sci.2023, 13, x FOR PEER REVIEW 9 of 30 quantifies how evenly the chemical differences are related to the retained components.
As can be seen in the compositional evenness graph, the elements that introduce more variability are MnO and CaO (tv/τ.j< 0.3) and, to a lesser extent, Na2O, Cu, K2O, Rb, and Zn (0.3 < tv/τ.j< 0.5), but no components introduce low variability (tv/τ.j> 0.9).Thus, total variation and compositional evenness indicate a complex structure in the data set.To summarise the data treatment, we present the results of the cluster analysis, using the squared Euclidean distance and the centroid agglomerative algorithm, performed on the clr transformed components: Na 2 O, MgO, Al 2 O 3 , SiO 2 , K 2 O, CaO, TiO 2 , V, Cr, MnO, Fe 2 O 3 , Ni, Cu, Zn, Ga, Rb, Sr, Y, Zr, Nb, Ba, Ce, Pb, and Th.The study of the dendrogram (Figure 5) indicates a complex structure with ten meaningful ceramic groups (or paste compositional reference units), defined as clusters of two or more individuals (plus group CGEIV, according to [22]), that account for just 33 individuals, plus 25 ungrouped individuals because of their compositional singularities in the present data set.A higher total variation should be envisaged for such a set of compositions, but the picture suggests a complex situation of different groups (   This structure can be understood with the biplots of the singular value decomposition on the double-centred clr transformed subcomposition Na 2 O, MgO, Al 2 O 3 , SiO 2 , K 2 O, CaO, TiO 2 , V, Cr, MnO, Fe 2 O 3 , Ni, Cu, Zn, Ga, Rb, Sr, Y, Zr, Nb, Ba, Ce, Pb, and Th (Figure 6) [42][43][44].A first glance reveals that the first principal component explains 49.22% of the variance and opposes the relative values of MgO, Cu, Zn, and, especially, MnO on the right-hand side and K 2 O and Rb on the left-hand one.The second principal component explains 18.13% of the variance and mainly reflects the attraction of CaO and Sr to the bottom of the biplot (Figure 6, top, left, and right).The main structure of the dendrogram (Figure 5) shows the existence of five individuals on the left, followed by two branches.The left branch corresponds to groups and individuals that exhibit high relative MnO, Zn, Cu, and MgO values, while the right branch corresponds to groups with high relative K 2 O and Rb values (Tables 2 and 3), as reflected by the first principal component.Additionally, groups and individuals are scattered from top to bottom in increasing relative values of Sr and, especially, CaO.If we draw an imaginary line from group CGGUA to group CGEIV, the individuals in the right-hand bottom corner of the biplots (Figure 6, top) correspond to those with CaO values over 9% (Tables 2 and 3), including four out of the five individuals between the two branches observed in the dendrogram (Figure 5) (MOS075, 045, 039, and 070), in the corner, and, on the imaginary line, MOS081 and MOS010, ungrouped individuals in the left-hand side branch.On the opposite top-left corner, we find group CG17 and individuals MOS053 and 019 with CaO concentrations of 2.05% or below (Tables 2  and 3), placed at the beginning of the right-hand side branch (Figure 5).Finally, the third principal component explains 8.83% of the variance and mainly reflects the increasing relative contents of Na 2 O from bottom to top (Figure 6, bottom).Thus, at the bottom, individual MOS045 has a Na 2 O content of 0.14%, while at the top, individual MOS051 has an extraordinarily high concentration of 2.79%.The latter is the fifth individual starting from the left on the dendrogram, just before the two branches.Besides that, the third component also shows the difference existing within the right-hand branch with group CG10 and individual MOS038 located in the upper part of the left side (Figure 6, bottom) because of their significantly higher Na 2 O relative values (Tables 2 and 3).In the following, we present an assessment of the chemical and mineralogical data obtained through the WD-XRF, PXRD, SEM-EDX, and PE analyses for each of these groups and several particular loners (Tables 3 and 4, and Supplementary Material 1: detailed petrographic descriptions).In the previous statistical treatments [22], it was possible to link some of the structures in the dendrogram resulting from the cluster analysis to production centres published by other researchers through the comparison with the existing reference groups.This comparison is semiquantitative, as only major and minor elements were considered, and there are no interlaboratory calibration studies.However, this starting point suggests possible provenances for some of our groups.Thus, it was possible to associate MOS069, 012, and 022 with Cerro del Villar (CGGUA), a Phoenician production centre in the mouth of the Guadalhorce river [45,46].Furthermore, we were able to relate MOS014, 062, 064, 080, 030, 006, and 007 with the site of Toscanos (CGTOS) in the Vélez river estuary, thanks to the study conducted at the Instituto di Ricerche Technologiche per la Ceramica-CNR (Faenza) on Phoenician tableware from this site [47,48].Moreover, our data were compared with those published by the Deutsches Archäologisches Institut, performed with neutron activation analysis [49,50].In this case, very few elements could be considered (Cr, Sc, Rb, La, Ce, Sm, Yb, and Fe 2 O 3 ).Despite this limitation, we observed how our group CGGUA matched their group 7, which was related to Cerro del Villar, and individuals MOS076, 077, 067, and 074 appeared to be related to their group 1, connected to the area around the River Vélez, where Phoenician production centres, such as La Pancha, have been excavated.For this reason, we decided to name this group CGVEL-MAG.Finally, we compared our results with the ARQUB database, relating specimen MOS050 to the reference group of Ses Figueretes [51], a production centre on the island of Ibiza during the third century BC.This compatibility suggests that a Phoenician production centre may have existed on the island during the Early Iron Age.Phoenician T.10.1.2.1.amphorae have been analysed from the site of Via Roma (northern Ibiza), dated to the 6th century BC, whose petrographic characterisation is compatible with the island's lithology and the petrographic group defined for the Ses Figueretes workshop [52].
In the following, we present an assessment of the chemical and mineralogical data obtained through the WD-XRF, PXRD, SEM-EDX, and PE analyses for each of these groups and several particular loners (Tables 3 and 4, and Supplementary Material 1: detailed petrographic descriptions).

Guadalhorce Valley CGGUA
This calcareous group includes two amphorae (MOS012 and MOS022) and one pithos (MOS069) that can be related to the El Cerro del Villar production centre, located in the Guadalhorce Valley in Málaga [45,49].The chemical data (Table 3) emphasise their significantly higher concentrations of Ni and Cr.
The petrographic study places the two studied individuals, MOS022 and 069, in petrographic Group 8 (Serpentinite and Well-Rounded Quartz) [22], in which the aplastic fabric component is dominated by rock fragments and the constituent minerals of metamorphic rocks (Figure 7, MOS069), where serpentinite and well-rounded quartz are the character-istic inclusions in both the coarse and fine fractions of a calcareous matrix.Carmona [53] has previously suggested the possible exploitation of the Mio-Pliocene clays in the valley's foothills as suitable raw materials for pottery manufacture.The Guadalhorce valley presents rich areas of sand and marl with schist, quartzite, amphibolites with staurolite, disthene, sillimanite, andalusite, and serpentinite outcrops.This geological background clearly matches the petrographic characteristics of this group.From the chemical and petrographic point of view, this group bears some similarities with group FG 3 in the work of Fantuzzi and collaborators [54].[55].Only in the case of CGTOS does a possible relation exist between the chemical group and the site of Toscanos, suggesting a specific, known provenance.For this reason, all samples related to the Vélez River area-several amphorae (MOS067, 074, and 076) and one pithos (MOS077)-but not ascribed to a specific workshop were labelled as CGVEL-MAG.Nevertheless, the significantly high CGVEL-MAG MnO concentrations separate the group from CGTOS (Table 3).The general link of the chemical composition of groups CG13 and CG14 and loners MOS010, 015, 024, 001 (amphorae), and 083 (dipper jug; see Figure 3j) with the Vélez River area should also be noted, though they vary sufficiently for their exclusion from the previous groups.For instance, group CG13 (MOS020 and 023) has high Cu and Zn concentrations (Table 3), and group CG14 (MOS013, 021, 061, 065, and 073) exhibits the highest Zn concentrations.These chemical similarities related to the Vélez-Málaga region were also detected by Behrendt and Mielke [49,50], suggesting a greater diversity in the production of this area [56].Even if they exhibit differences, loners MOS010, 015, 024, 001, and 083 are grouped as CG15 to point to an origin in a similar geological environment, but it is important to keep in mind that this group is heterogeneous and does not seem to represent a specific production unit or location.PXRD allows the identification of two different fabrics after the association of crystalline phases [35] within this calcareous group (CaO 10.45% in normalised data): CGGUA-I, represented by sample MOS069, and CGGUA-II, represented by samples MOS012 and 022 (Table 4).In the case of CGGUA-I, it is possible to estimate an equivalent firing temperature (EFT) of around (950-1000) • C on account of the absence in its diffractogram of the d (002) peak at 10 Å of illite-muscovite, and the presence of clear firing phases like gehlenite.Fabric CGGUA-II shows the total decomposition of illite-muscovite and gehlenite, and its EFT should be around (1000-1050) • C.This analysis confirms a differentiation observed through PE, where the micromass is optically inactive due to a possible high firing temperature.
SEM-EDX analysis of fabric CGGUA-I reveals continuous vitrification (Vc), confirming an EFT between (850-1050) • C (Figure 8, sample MOS069), the same inference that can be made for fabric CGGUA-II (Figure 8, sample MOS012).PE analysis shows that the main characteristic of CGTOS (including individual MOS009) and CGVEL-MAG ceramics, all classed in Group 9, is the presence of metamorphic rocks as the predominant non-plastic inclusion in a dense calcareous matrix (Figure 7, MOS009).Fragments of schist and rounded to well-rounded elongate phyllites dominate the coarse fraction, while muscovite mica, quartz, albite, and opaques indicate lowgrade metamorphism.There may also be sedimentary rocks that have been changed by metamorphism, amphiboles, garnet, and hornblende that have been changed into serpentine.These petrographic characteristics are also present in groups CG13 and CG14, indicating a nearby provenance.The similarities between the geological formations of the Riff area (Morocco) and the Baetic mountain range [57] hinder a more precise determination of provenance.Previous PE studies of Phoenician ceramics found at Ceuta [58], on the African side of the Strait of Gibraltar, suggested a distinction between these two areas, contrasting the presence of ultrabasic rocks, a possible indicator of a Morocco provenance, with a low grade of metamorphics, linked to an Andalusian origin.In our case, it is not possible to suggest or reject a Moroccan provenance for groups CG13 or CG14 until we have further data from known production centres.All these groups exhibit some similarities with groups FG 4-5 [54], and, remarkably, group CG14 exhibits high Al2O3 and Zn concentrations.

Toscanos CGTOS and Vélez Valley CGVEL-MAG, with Groups CG13 and CG14
Individuals MOS006, 007, 009 (at the left of the group in the dendrogram of Figure 5), 014, 062, 064, 080 (corresponding to amphorae) and MOS030 (a tripod mortar/vessel) were related in previous statistical treatments [22] to the reference group of the Phoenician site of Toscanos [47,48].The principal chemical differences between CGGUA and CGTOS are the lower Cr and Ni concentrations in the latter, besides having the lower MnO among the ceramics in the left-hand branch of the dendrogram (Table 3).They also have different petrographic characteristics where, in contrast to the presence of serpentinite in the Guadalhorce ceramics, mica-schist is the main inclusion characterising products of the Vélez valley in what is considered Group 9 (Metamorphic Rocks).However, the distinction between different groups deriving from the area of the Vélez River is challenging, both through mineralogical and chemical means [49,50].Archaeological remains from several workshops in the area reflect the intensity of ceramic manufacture during the Iron Age [55].Only in the case of CGTOS does a possible relation exist between the chemical group and the site of Toscanos, suggesting a specific, known provenance.For this reason, all samples related to the Vélez River area-several amphorae (MOS067, 074, and 076) and one pithos (MOS077)-but not ascribed to a specific workshop were labelled as CGVEL-MAG.Nevertheless, the significantly high CGVEL-MAG MnO concentrations separate the group from CGTOS (Table 3).The general link of the chemical composition of groups CG13 and CG14 and loners MOS010, 015, 024, 001 (amphorae), and 083 (dipper jug; see Figure 3j) with the Vélez River area should also be noted, though they vary sufficiently for their exclusion from the previous groups.For instance, group CG13 (MOS020 and 023) has high Cu and Zn concentrations (Table 3), and group CG14 (MOS013, 021, 061, 065, and 073) exhibits the highest Zn concentrations.These chemical similarities related to the Vélez-Málaga region were also detected by Behrendt and Mielke [49,50], suggesting a greater diversity in the production of this area [56].Even if they exhibit differences, loners MOS010, 015, 024, 001, and 083 are grouped as CG15 to point to an origin in a similar geological environment, but it is important to keep in mind that this group is heterogeneous and does not seem to represent a specific production unit or location.
PE analysis shows that the main characteristic of CGTOS (including individual MOS009) and CGVEL-MAG ceramics, all classed in Group 9, is the presence of metamorphic rocks as the predominant non-plastic inclusion in a dense calcareous matrix (Figure 7, MOS009).Fragments of schist and rounded to well-rounded elongate phyllites dominate the coarse fraction, while muscovite mica, quartz, albite, and opaques indicate low-grade metamorphism.There may also be sedimentary rocks that have been changed by metamorphism, amphiboles, garnet, and hornblende that have been changed into serpentine.These petrographic characteristics are also present in groups CG13 and CG14, indicating a nearby provenance.The similarities between the geological formations of the Riff area (Morocco) and the Baetic mountain range [57] hinder a more precise determination of provenance.Previous PE studies of Phoenician ceramics found at Ceuta [58], on the African side of the Strait of Gibraltar, suggested a distinction between these two areas, contrasting the presence of ultrabasic rocks, a possible indicator of a Morocco provenance, with a low grade of metamorphics, linked to an Andalusian origin.In our case, it is not possible to suggest or reject a Moroccan provenance for groups CG13 or CG14 until we have further data from known production centres.All these groups exhibit some similarities with groups FG 4-5 [54], and, remarkably, group CG14 exhibits high Al 2 O 3 and Zn concentrations.
(850/900-950/1000) °C.Finally, group CG14 has two fabrics, CG14-I (MOS013, 021, and 065) and CG14-II (MOS061 and 073), whose main difference is the presence or absence of illite-muscovite, providing two different ranges of EFT (850/900-1000) °C and over 1000 °C, respectively.The EFT estimated for CG14-II is sharpened with the information inferred from SEM-EDX (Figure 8, MOS061), which shows a Vc+ state corresponding to the range (1050-1080) °C.It is important to point out the presence of andalusite and garnets in these groups.PXRD of VEL-MAG reveals three different fabrics (Table 4).CGVEL-MAG-I (MOS076 and 077) has an EFT of 800-850 • C if hematite is considered a firing phase, but no clear firing phase can be observed.Andalusite, quartz, plagioclase, and illite-muscovite are also present.Fabric CGVEL-MAG-II (MOS074) is similar but exhibits spessartine, and its EFT should also be in the range of 800-850 • C. Finally, fabric CGVEL-MAG-III (MOS067) displays the absence of illite-muscovite and calcite and the presence of quartz, hematite, pyroxene, and plagioclase (Figure 9, top left).This suggests that the EFT was over 950-1000 • C. In the case of CGVEL-MAG, no SEM-EDX analysis was performed.
Group CG13 exhibits only one fabric (Table 4), CG13-I (MOS020 and 023), that, after the PXRD and SEM-EDX pieces of evidence (Figure 8, MOS023), suggests an EFT of (850/900-950/ 1000) • C. Finally, group CG14 has two fabrics, CG14-I (MOS013, 021, and 065) and CG14-II (MOS061 and 073), whose main difference is the presence or absence of illite-muscovite, providing two different ranges of EFT (850/900-1000) • C and over 1000 • C, respectively.The EFT estimated for CG14-II is sharpened with the information inferred from SEM-EDX (Figure 8, MOS061), which shows a Vc+ state corresponding to the range (1050-1080) • C. It is important to point out the presence of andalusite and garnets in these groups.

Ibiza: CGEIV
Based on the chemical composition, sample MOS050 is related to the Punic production centre of Ses Figueretes (Ibiza), previously studied by our group [51].This correlation bolsters the suggestion of a Phoenician workshop in operation on the island during the Early Iron Age.Previous studies have identified such products on the Catalan coast as amphorae from Sant Martí d'Empúries.Macroscopic observations linked them to Ibiza [60], and chemical analysis confirmed this hypothesis [61].However, pottery from Ibiza is scarce on the Catalan coast compared to that of Andalusian vessels, which are the dominant source of the whole assemblage [22].
The island of Ibiza is dominated by sedimentary and calcareous formations, mainly limestone, dolostone, and marl outcrops [51,62].PE examination of individual DOV002 (not included in the present study; see [22]), an amphora recovered at the site of Aldovesta that is chemically related to MOS050 and therefore to group CGEIV and Group 3 (Alotriomorph Quartz and Mica Matrix) [22], suggests that this group is characterised by very rounded quartz and mica inclusions throughout its matrix.This amphora from Aldovesta (Figure 7, DOV002) contains micritic calcite with merging boundaries as the coarse fraction's dominant component, while the frequency of muscovite mica in the fine fraction stands out.Also characteristic of this fabric are clay pellets, dark orange in PPL and brown in XP (x25), with clear boundaries and high optical density, generally sub-angular and discordant with the micromass, and fine-grained non-plastic inclusions of monocrystalline quartz and calcite.
The PXRD analysis of this calcareous fabric, CGEIV-I (18.01%CaO in normalised data, Table 3), shows the presence of quartz, plagioclase, pyroxene, gehlenite, and calcite, the latter of possibly primary and secondary origin (Table 4).The absence of illite-muscovite and the presence of gehlenite suggest a high EFT in the range of (1000-1050) • C. No SEM-EDX analysis was performed on individual MOS050.

Groups without a Defined Provenance: CG21, CG10, CG17, CG19
A clear provenance could not be suggested for groups CG21, CG10, CG17, and CG19, which, together with CGEIV and twelve loners (MOS053, 019, 038, 047, 008, 071, 066, 072, 052, 049, 027, and 033), comprise the right branch of the dendrogram (Figure 5) and include most of this study's jars and tripod mortars.In contrast, the left branch of the dendrogram includes most of the amphorae.The exception is group CG10, comprised only of amphorae, whose provenance is still uncertain.

Group CG21
Group CG21 includes two mortars (MOS025 and 048).MOS048 is included in Group 4 (Sandstone Inclusions and Serpentinite), which reveals common sandstone fragments in the coarse and fine fractions, together with opaques and quartz (Figure 7, MOS048).This group is calcareous (6.6% and 6.37% in normalised data), and PXRD shows the presence of illite-muscovite, calcite, quartz, plagioclase, and K-feldspar phases in both individuals, indicating an EFT below (800/850) • C. Analysis of MOS048 using SEM-EDX confirmed an initial vitrification stage, confirming an EFT in the range of (750-800) • C (Figure 8, MOS048).

Group CG10
The five amphorae (MOS005, 011, 016, 029, and 032) of the borderline-calcareous group CG10 present some shared, characteristic morphological features, being atypical of the T-10.1.2.1 type.It can be suggested that they are not of the Phoenician type (J.Ramon personal communication), and they have been labelled as T.2.1.1.2or similar (Table 1).The rim exhibits a much more everted shape, and they do not have the expected "baggy" body (Figure 3a1).Their provenance remains uncertain since the group's chemical composition does not match any available reference group.The PE analysis shows the presence of metamorphic and igneous rock fragments in individuals MOS029 and MOS032 conforming to Group 5 (Metamorphic and Igneous Rock), which is characterised by dominant, coarse, angular to sub-angular basic igneous rocks, some with possible trachytic texture and a distinctive yellowish to reddish mineral (Figure 7, MOS032).This phenomenon is repeated in most igneous rocks, possibly because of alteration and minerals altered to serpentine.Rounded iron oxides of red to brown colour and possible epidotes, with a hexagonal shape in MOS029, are common in the fine fraction.Elongate and well-rounded opaques with clear boundaries, angular volcanic glass with sharp boundaries, and acicular crystals are also present.There are very few to very few sub-angular to sub-rounded clinopyroxene and small monocrystalline quartz that is sub-rounded to well-rounded.Although the provenance of CG10 remains uncertain, it must be stated that intrusive igneous rocks are attested throughout the lower Guadalquivir valley along the western margins of the Betic Cordillera [63,64], but further research is required.PXRD analysis of these bordercalcareous amphorae (4.44% of CaO in normalised data, Table 3) shows two different fabrics (Table 4).CG10-I (MOS005 and 016) has illite-muscovite, K-feldspar, hematite, plagioclase, and spinel as a firing phase with an EFT of ca.(900-950) • C. For CG10-II (MOS011, 029, and 032), the d (002) peak at 10 Å of illite-muscovite is no longer observable, allowing us to fix an EFT between (950-1000) • C. Individuals MOS005 (CG10-I) and MOS032 (CG10-II) were studied by SEM-EDX and displayed extensive vitrification, which enabled us to confirm the PXRD EFT values (Figure 8, MOS005 and 032).

Group CG17
Group CG17 contains an amphora (MOS063) and a jar (MOS068), whose chemistry does not allow the ascription of provenance.PE of MOS063 shows the presence of textural concentration features or opaques as the predominant non-plastic inclusion and corresponds to Group 6 (Textural Concentration Features/Grog Tempered).These features, observed in both the coarse and fine fractions throughout the matrix, are generally dark red to opaque in PPL and brown-reddish to opaque in XP (x25), with sharp to clear boundaries and very high optical density (Figure 7, MOS063).Generally, they are very angular to subangular and discordant with the micromass.Sub-rounded calcite crystals and well-rounded micrite are dominant and frequent in the coarse fraction.Monocrystalline angular to wellrounded quartz is frequent in the fine fraction and white mica crystals.There is evidence of the mixing of two different clays in some parts of the thin section.PXRD analysis of these low-calcareous vessels (1.22% of CaO in normalised data, Table 3) shows a unique fabric (CG17-I), which presents a low EFT of <(800/850) • C since no firing phases are observed (Figure 9, bottom middle).Individual MOS068 was studied by SEM-EDX, exhibiting initial vitrification of the matrix (Figure 8, MOS068) that suggests an EFT estimate of (750-800) • C.

Group CG19
Group CG19 (two jars, MOS054 and 079) cannot be ascribed to a specific provenance.In fact, MOS079 is included in Group 4 (Sandstone Inclusions and Serpentinite), which does not contain mica-schist inclusions, suggesting a provenance other than the area of Granada and Málaga (Figure 7, MOS079).PXRD analysis of these low-calcareous ceramics (3.85% of CaO in normalised data, Table 3) produced one fabric, CG19-I, where the d (002) peak at 10 Å of illite-muscovite is no longer present and which contains quartz, hematite, spinel, and plagioclase phases, suggesting an EFT around (950/1000) • C (Figure 9, top middle).The SEM-EDX of MOS054 shows extensive vitrification of the matrix, confirming the EFT indicated by PXRD (Figure 8, MOS054).

Loners Possibly Related to These Groups
Twelve single individuals or loners (MOS053, 019, 038, 047, 008, 071, 066, 072, 052, 049, 027, and 033) (Figure 5) are possibly related to the above groups, except for group CG10, which stands clearly apart.The loners share general similarities with groups CG17, CG19, and CG21, even if several differences in their chemical composition prevent their classification in those groups.Several aspects are worthy of consideration.Individuals MOS066, 072, and 027 are classified into petrographic Group 4, as is the case with the members of chemical groups CG19 and CG21.Individuals MOS0039 and 070, on the left-hand side of the dendrogram (Figure 5), are also classified into petrographic Group 4, but their chemical composition shows more significant differences, notably much higher CaO and Sr contents.Individual MOS047 is classified into petrographic Group 6, as is individual MOS063 from chemical group CG17.
In addition to the petrographic evidence, in chemical terms, it is significant that, even if Sn was disregarded for the data treatment on account of several values below the lower limit of regression, some individuals exhibited atypically high Sn concentrations that might be diagnostic in the future for assigning provenance.Individuals MOS054 and 079 (CG19), MOS068 (CG17), and loners MOS053 and 019 have concentrations from 52 to 107 mg/kg, while loner MOS038 goes up to 266 mg/kg (Table 2).Such concentrations might be related to areas of tin mineralisation.The Iberian Peninsula is rich in such mineralisations that have been exploited since the Bronze Age, especially in the Hesperian (or Iberian) Massif, which covers most of the western half of the Iberian Peninsula.The northwestern part of this Massif is particularly rich in such tin mineralisations forming the Iberian tin belt [65].However, the northwest of the Iberian Peninsula, where this tin belt reaches the coast, has no evidence of western Phoenician pottery production.Within the southern part of the Hesperian Massif, two units of interest are the South-Portuguese Zone, at the southernmost side of the Massif, and the Ossa-Morena Zone, just to its north.The South-Portuguese Zone contains the Iberian Pyrite Belt (IPB), one of the largest mining districts in the world, including Sn mineralisations, for example, in the Alentejo [66].This unit occupies the area from the province of Sevilla and south of Lisbon, almost to the Atlantic coast.In the search for tin, among other metals, the Phoenicians established several colonies on that coast, especially in the Tagus estuary, where they produced ceramics (e.g., [67,68]).Within the Ossa-Morena Zone, there is evidence of tin mining and metallurgy at Cerro de San Cristóbal (Logrosán, Cáceres, central Spain) since the Late Bronze Age [65], but also some tin ingots from the Phoenician shipwreck of del Bajo de la Campana (Cartagena, Murcia) (ca.625-575 BC) have an origin in Los Pedroches (Córdoba) [69,70].Finally, tin mineralisation also exists in the area of La Unión, in the coastal area of Murcia at Portman Bay [71], in the southeast of the Iberian Peninsula.However, again, there is no evidence of western Phoenician pottery production in the vicinity.

Unrelated Loners
Individuals MOS075, 045, 051, 081, and 078, at the left-hand side of the dendrogram (Figure 5), appear as loners with no relation to any defined groups.Their chemical compositions differ significantly from all other individuals and the ceramics in comparative studies.Nevertheless, two individuals must be commented on in some detail.Individual MOS051 exhibits particularly high Na 2 O (2.79%) together with high K 2 O (4.39%).By PXRD, the most intense peak corresponds to plagioclase, possibly albite, and it also exhibits very intense peaks of K-feldspar, pyroxene, and analcime.Quartz and illite-muscovite are also present (Figure 9, right).Unfortunately, this individual could not be studied by PE, but all the identified phases could be considered primary phases, pointing to a particular geological environment probably related to vulcanism.Individual MOS075 was classified into PE Group 1 (Rounded Quartz and Calcite Inclusions), which is related to Central Mediterranean Phoenician imports already described elsewhere [23].However, despite its petrographic comparisons, chemical analysis reveals significant differences, though it cannot at present assign a particular provenance.

Discussion
Analysis of the pottery assemblage from Sant Jaume has produced compositional groups with some confidence, of which we have been able to suggest a provenance for six.The trends revealed from these ceramics, thought to be associated with Phoenician settlement and activities in the western Mediterranean, reveal some clear patterns in terms of space and time.Considering loners as potentially different groups (except those gathered together in the so-called group CG15), we have defined29 groups of wheel-made pottery, complemented by 3 groups of central Mediterranean Phoenician imports [23].Six groups plus CG15, accounting for 28 individuals, originated in the Western Mediterranean Phoenician colonies of the southeast of the Iberian Peninsula, in the coastal areas of Málaga, Granada, and Almería, or the northern coast of Morocco (CGGUA, CGTOS, CGVEL-MAG, CG13, CG14, and CG15), and the island of Ibiza (CGEIV).The other 30 individuals account for the remaining 23 groups (groups CG21, CG10, CG17, and CG19, and 19 loners).The observed diversity in provenance provides further evidence of a complex commercial network involving a large number of pottery workshops from different geographical areas.In Figure 10, we can see that diversity is very high, with richness equal to 29, the number of groups defined, while evenness, the distribution of individuals in each group, is considered very high (H 2 = 4.45 Sh, i.e., 91.63% of the total attainable).The high value of evenness reflects the absence of preferential consumption of goods from a single production centre, or even a small number of sources, emphasising the complex dynamic between a variety of centres, with no clear predominance of any particular Phoenician colony.
nician colonies of the southeast of the Iberian Peninsula, in the coastal areas of Málaga, Granada, and Almería, or the northern coast of Morocco (CGGUA, CGTOS, CGVEL-MAG, CG13, CG14, and CG15), and the island of Ibiza (CGEIV).The other 30 individuals account for the remaining 23 groups (groups CG21, CG10, CG17, and CG19, and 19 loners).The observed diversity in provenance provides further evidence of a complex commercial network involving a large number of pottery workshops from different geographical areas.In Figure 10, we can see that diversity is very high, with richness equal to 29, the number of groups defined, while evenness, the distribution of individuals in each group, is considered very high (H2 = 4.45 Sh, i.e., 91.63% of the total attainable).The high value of evenness reflects the absence of preferential consumption of goods from a single production centre, or even a small number of sources, emphasising the complex dynamic between a variety of centres, with no clear predominance of any particular Phoenician colony.There are clear indications of a large number of workshops in the south of the Iberian Peninsula that manufactured mainly amphorae; certainly, this area was the provenance most represented in transport jars at Sant Jaume, with the only significant exception being the CG10 group.It is difficult to determine more specific provenance for groups CG13 and CG14 and the outliers grouped within CG15 in that area of Andalusia (especially Málaga, Granada, and Almería provinces), where archaic Phoenician colonies were located, due to the occurence of the same geological units with similar compositional characteristics.Similarly, the geology of Morocco on its northeastern Mediterranean coast also shares similarities to that of Málaga/Granada, a problem not helped by the lack of analytical studies of pottery of this time period in Morocco, implying that we cannot rule out a north-African provenance for CG13 and CG14.Nevertheless, it has been possible to make clear distinctions between the Guadalhorce Valley (West Málaga, group CGGUA) and the Vélez Valley (East Málaga, group CGVEL-MAG) through chemical and mineralogical analyses, as well as to identify a group probably related to the Phoenician site of Toscanos (CGTOS).Although recent petrographic and chemical studies of Phoenician pottery have been carried out in Andalusia [64,72,73], we are still far from having a complete picture.Future large analytical projects on the assemblages of Phoenician colonial sites are necessary to enhance the characterization of production units and provide crucial comparative data.
Nevertheless, some interesting differences appear if we compare the consumption patterns revealed by the analyses with those of later periods.Although we have identified the importation of pottery vessels from Ibiza (group CGEIV), these are relatively rare in this early phase on the Catalan coast.This pattern changed drastically during the Punic period [74] when Ebussitan vessels were common in Catalan-Iberian sites.
While it was not possible to suggest a provenance for most of the tripod mortars/vessels, they display great diversity.Most are not related to the area of Málaga or Granada, and two (MOS025 and 048) are members of the same group (CG21) of unknown provenance.Only MOS030 is possibly associated with the T-10.1.2.1 amphorae from Toscanos (CGTOS).It is also clear that the narrow-necked cylindrical jars, or Cruz del Negro type, do not share a provenance with that suggested for most of the amphorae.A possible provenance in Gadir and the surroundings of the Bay of Cádiz has also been ruled out, as the group does not correspond to a reference group from Camposoto (unpublished data available in our databank) or the available materials studied by other scholars.In this case, as with CG10, it has not been possible to suggest a provenance with confidence, although we can certainly say that the amphorae in group CG10 do not have their origin in the area of Málaga or Granada, an observation compatible with their variations in morphology from T-10.1.2.1 amphorae.
Previous studies have pointed out how consumption patterns among Phoenician sites in Andalusia show that the exchange between the Atlantic and Mediterranean regions was very limited [56].In our results, even though the western Mediterranean region has a prominent role and seems to dominate the commercial network up to the northeast of the Iberian Peninsula, it is essential to keep in mind that we cannot point to a region of origin for a majority of the studied ceramics (30 out of 58).Despite this problem of sourcing the ceramics, the diversity detected would suggest it is a pattern resulting from cabotage navigation and not an exchange directed by any one Phoenician colony.Similar behaviour has been detected in the metal trade by analytical studies in the northeast of the Iberian Peninsula [75].
While we have show the emergence of important patterns of provenance for Phoenician pottery, it is equally clear that further chemical, mineralogical, and petrographic work would be beneficial beyond the Guadalhorce and Vélez Valleys and particularly on specific ceramic assemblages not only in the areas of Almería, Murcia, and Alacant but also in Morocco and on the Atlantic coast of Huelva and Portugal.

Figure 1 .
Figure 1.Map of the locations of Phoenician colonies in Ibiza (a), on the southeastern coast of the Iberian Peninsula (b), and the Early Iron Age reception sites of Phoenician pottery along the northeastern coast (c).

Figure 1 .
Figure 1.Map of the locations of Phoenician colonies in Ibiza (a), on the southeastern coast of the Iberian Peninsula (b), and the Early Iron Age reception sites of Phoenician pottery along the northeastern coast (c).

Figure 2 .
Figure 2. General plan of Sant Jaume.Figure 2. General plan of Sant Jaume.

Figure 2 .
Figure 2. General plan of Sant Jaume.Figure 2. General plan of Sant Jaume.

Figure 4 .
Figure 4. Compositional evenness graph of the 58 studied individuals.H2: information entropy (in Shanons, Sh); H2 %: percentage of the maximum possible attainable; tv: total variation.τ.j: trace of the variance-covariance matrix following the alr transformation using element j as the divisor.Vertical dotted lines express different tv/τ.jvalues.

Figure 4 .
Figure 4. Compositional evenness graph of the 58 studied individuals.H 2 : information entropy (in Shanons, Sh); H 2 %: percentage of the maximum possible attainable; tv: total variation.τ .j: trace of the variance-covariance matrix following the alr transformation using element j as the divisor.Vertical dotted lines express different tv/τ .jvalues.

Table 3 .
Mean (X), standard deviation (s), and total variation (tv) of the groups of more than two individuals and the values of individual MOS050 in group CGEIV (as normalised values).Major and minor elements (expressed as oxides) in w%.Trace elements in w mg/kg.

Table 2 .
Chemical concentrations by WD-XRF.Major and minor elements expressed as oxides are in w%.Trace elements in w mg/kg.LOI (loss on ignition) in w%.b.l.d.: below the limit of detection.

Table 2 .
Chemical concentrations by WD-XRF.Major and minor elements expressed as oxides are in w%.Trace elements in w mg/kg.LOI (loss on ignition) in w%.b.l.d.: below the limit of detection.

Table 3 )
sharing similar geological backgrounds in some cases. s

Table 4 .
Summary of the mineralogical fabrics defined by PXRD analysis for each meaningful chemical group.n: number of individuals.EFT: equivalent firing temperature. PE