Some Notes on Dense Structures Present in Archaeological Plant Remains: X-ray Fluorescence Computed Tomography Applications

Abstract

This study describes the composition and provenance of dense bodies or structures present in uncharred plant vestiges recovered at Monte Castelo (6000–700 cal. AP-SW Amazonia). It aimed to disclose some aspects of this plant remains’ interactions with the sedimentary matrix of the site over the 200 years (at least) since its initial deposit, from the point of view of the soft tissue mineralization processes. Two specimens were examined using XFCT, X-ray MicroCT, and SEM-EDS techniques to reveal the presence and distribution of Ca, K, Mn, Fe, Ti, Si, S, Cu, Br, Rb, Sr, Zn, and Zr. These attributes were integrated with compositional ED-XRF and XRD measured data from the sedimentary substrate. Results show that some of the chemical elements present in solid bodies and anatomical structures of the plant remains refer to the sedimentary environment, while others have an endogenous origin. These include mainly Rb and Br, which were interpreted as the result of degradation processes of the internal tissues, where they are mainly present. Except Sr and Zr, a portion of all the other elements entered and disperse into the sample structures from the sedimentary substrate. Its presence is attributable to mechanisms such as attachment, diffusion and impregnation through the outermost tissues, where they are mostly concentrated. The composition of most of the dense bodies consists of both endogenous and exogenous elements.

1. Introduction

This article builds on some assumptions derived from previous X-ray microtomography analysis of a set of uncharred plant remains from the archaeological site Monte Castelo (Rondonia, Brazil), a fluvial shellmound in Southwestern Amazonia [1,2,3]. Tridimensional images of the internal structure of these specimens had shown the presence of solid aggregates, or concretions, whose relative density, distribution, and shape does not refer to anatomical features [4,5]. A mineral composition derived from some mechanical, biological, and/or physicochemical interaction processes involving the sedimentary materials from the site matrix was then proposed as the origin of these dense bodies.

The presence of exogenous minerals coating and permeating archaeobotanical specimens has frequently been reported for sites where specific environmental/contextual characteristics allowed mineralization, principally the abundance of dissolved ions and some fluctuating presence of water [6,7,8]. Mineralization involves the gradual replacement of biological tissues with minerals which can naturally occur at different stages of the remains’ deposition and early diagenesis. It exhibits several modes according to the internal variations of the specimen and environmental differences present in the deposits [9]. For example, phosphate mineralization is one of the most studied mineralization processes in archaeobotanical remains. The abundance of Ca and P compounds in the site matrix, along with the fluctuant presence of a liquid medium, allows these ions to diffuse and concentrate into the plant specimens. There, they will combine and precipitate as calcium phosphate compounds, which gradually replace other minerals in the decaying remain [6,10,11,12].

Mineralization has been observed as a taphonomic process, mostly enhancing the preservation of non-charred organic remains in open sites from wet environments with increased alkalinity (high pH values) or acidity (low pH values) of the soils [13]. It has been also studied as a potentially distortive agent for radiocarbon dating of archaeobotanical material [14]. Compounds with Cu, Fe, and phosphates have already been observed in coatings that prevent the progression of various degradation mechanisms in plant remains [7,10,11,12,15,16]. Other mineral substances, such as silicates, carbonates, gypsum, and calcite, might also have fulfilled similar performances when present in the sedimentary substrate and/or other items in ancient deposits [13].

The abundance of Ca—as well as P, Mg, K, and Mn—compounds in Monte Castelo, mostly due to the decay of high amounts of shell material, and the fluctuating water levels along the seasonal regime of the local wetlands landscape, give the general conditions for the occurrence of some mineralization processes [1,3,5]. However, these have been minimally studied so far, especially in the case of preserved uncharred plant remains. This is partly due to the fact that the radiocarbon dating of some of these findings revealed uniformly recent dates, between 1820 and 1920 AD [3,4], trivializing the significance of any in situ preservation process that may have taken place.

Possible alterations in this recent radiocarbon dates due to contamination of the samples with modern charcoal still need to be examined. However, it should be noted that the older the age calculated for this material, the more likely it will be associated with processes other than carbonization to facilitate their preservation at the site, at least over a period of 100–200 years. This process could relate to the gradual acquisition and stay of different mineral compounds inside the plant remains from the enriched sediments at Monte Castelo. It is proposed that the observed solid bodies and other potential features relate to initial stages of this interaction process involving the compositional characteristics of the substrate and those of the plant specimens.

Considering the currently available data, this work is intended firstly to recognize the presence and distribution of different chemical elements through anatomical and non anatomical (dense bodies) structures of some uncharred plant remain samples. The analytical techniques X-ray Fluorescence Computed Tomography (XFCT) and Energy Dispersive X-ray Fluorescence Spectrometry (ED-XRF) were used for this purpose. These results were then compared with data on the compositional characteristics of the sedimentary substrate, in order to infer their provenance and some interaction mechanisms related to the acquisition and organization of the chemical elements. Compositional analysis of the sedimentary samples was performed using Energy dispersive X-ray Fluorescence (ED-XRF) and X-ray Diffraction (XRD) techniques.

Although ED-XRF, XRD, and SEM-EDS techniques have an extensive application in archaeological organic material analysis, and others such as X-ray absorption Microtomography are currently undergoing an intensive exploration of their potential, XFCT has no precedent in this field of study. This non-destructive imaging technique is able to reveal structural 3D inner details of the objects with high resolution and elemental specificity. While the analysis of trace elements distribution using XRF microscopy has been applied to the study of biological materials, i.e., [17,18,19,20], the development of XRF tomography is expanding the possibilities for structural visualization in biological samples, especially after several technical advances, overcoming their most significant previous limitations in terms of acquisition time, resolution range, reconstruction process, and analytical complexities, i.e., [21,22,23,24,25].

The importance of understanding the interactions between archaeological artifacts and their specific natural environment was pointed out early on by Schiffer, in his studies concerning cultural and non-cultural processes shaping the formation of archaeological sites. These interactions introduce variability in the archaeological record and affect our inferences about the past [26]. Since then, the range of archaeological artifacts, materials, and variables considered for study has consistently expanded to include different microparticles, molecular substances, and other chemical compounds and elements. Additionally, the techniques employed to characterize these objects and track the behavior of multiples physicochemical variables have evolved in more sophisticated measurement techniques and instruments, enhancing the detection of subtle variations [27].

The interactions between artifacts and environment persist and diversify at this microarchaeological and archaeometric order and the variability they produce must also be recognized and advantageously integrated into the analysis of archaeological materials. Far from hindering or disabling interpretations of artifacts and contexts, this kind of information tends to strengthen them, assist in exploring data bias and gaps, and fine tuning the analytic methodologies. Accordingly, this study seeks to contribute to a more general and comprehensive model of the physicochemical changes occurring in artifacts and materials from archaeological deposits within environmental and structural patterns comparable with those at Monte Castelo.

2. Archaeological Context: Soil, Shells, and Plant Remains in Monte Castelo

The archaeological site Monte Castelo is located on the floodplains of the middle Guapore River, in the Southwest Brazilian Amazon (Figure 1A). The site is a 6.3 m high shellmound covering an area of ca. 120 m by 100 m. It was built in an ecologically transitional area between the seasonally flooded savannas and the evergreen tropical forest. It consists of at least 21 archaeological strata combining gastropod shells and natural and anthropogenic soils as the result of a largely uninterrupted occupation sequence between ca. 6000 and 700 cal. BP [1,3,28,29,30]. This study examines some aspects of the most recent stratigraphic assemblage from Layer A to Layer D. The recovery of large quantities of ceramics and faunal remains characterizes the Bacabal strata (Layers D to B), which are dated between 4300 and 2000 cal. BP. Meanwhile, Layer A (700 cal. BP) also contains remains from a farm present atop the site between 1970 and 1980, as well as the remnants of temporary camps made by indigenous groups that still use the site as a hunting spot today (Figure 1B) [1,2,30,31,32,33].

The surface Layer A (10 to 20 cm thick) is composed principally of humus and soil litter. It contains some ceramic and lithic materials, but shells are very scarce. The Layer B (30 cm thick) presents a brown lime-clay-sandy matrix with high amounts of shells and abundant fragments of decorated Bacabal ceramic. In the Layer C (30 cm thick), the shell content is lower but combined with a high concentration of Bacabal ceramic fragments. It also presents a sub-adult burial (Burial 1) lying in a cavity delimited by gastropod shells. Some preserved activities areas were found as well. The Layer D (30 to 60 cm thick) is formed by clay-silt-sandy sediments becoming progressively blackened towards the base of the layer. Several lateritic granules and pebbles are present, and some portions of the soil are completely concretized. Ceramic is scarce, but some of the fragments have soot vestiges. High amounts of broken shells and preserved activity areas were found, as well as an adult burial (Burial 2), also decorated with shells and a deer antler [1,2].

Previous XRF analysis of the sedimentary substrate of the Bacabal layers defines them as classic Anthropic Dark Soils (ADS) or ‘terra preta’, containing high amounts of CaO (26.32% to 38.54%) and P₂O₅ (15.83% to 25.91%) combined with MgO (0.30% to 0.47%), K₂O (0.30% to 0.65%), MnO (0.14% to 0.26%), and organic matter (14.59% to 18.13%), directly associated with human occupation and the use of fresh water malacological fauna. In addition, abundant quantities of Fe₂O₃ (1.18% to 2.27%) were found, especially in layers A, B, and C, but they are proportionally lower than those present in off-site sediments. On the other hand, SiO₂ (17.57% to 35.52%) and Al₂O₃ (1.95% to 3.85%) were found in low quantities when compared with off-site areas. These elements were added to the site from the lake sediments enriched with the siliceous spicules of sponge animals and were used to construct the mound [1,2].

A considerable large and morphologically homogeneous assemblage of uncharred plant remains was recovered from the ceramic layers from the Bacabal Phase and the more ancient Sinimbu Phase (Figure 1C). They were more frequently found in the sediments of the layers C to E and particularly concentrated in association with the sub-adult burial context covering a sector of layer C [3,34]. Archaeobotanical analysis describes these plant remains as small non-charred drupaceous fruits related to the Anacardiaceae family (Figure 1D). In general, the specimens measure approximately 5 mm in maximum length and diameter. They present a three-layered pericarp with an outer layer or exocarp covering the mesocarp in the middle position, and the endocarp tissue which protects the seed and embryo tissues. Microtomographic images obtained from a dozen of these specimens revealed the occurrence of dense aggregates stuck into the internal tissues and cavities of the drupes. Some of them are regular small crystal bodies associated with the endocarp idioblast cells, while others were described as irregular, bigger, and apparently amorphous structures, placed between the mesocarp cells and inside some cavities in the fruit structure [4,5].

3. Materials and Methods

Two uncharred drupe specimens from Monte Castelo (SMC_01 and SMC_11) were analyzed. The sample SMC_01 found in association with the sub-adult burial context was imaged twice, using the XFCT technique at the XRF Beamline of the Brazilian Synchrotron Light Laboratory (LNLS-CNPEM) (UVX source), aiming to obtain compositional and distributional information about the solid bodies and anatomical structures of the fruit. Both experiments were set to provide XFCT images of the internal structure of the sample that can be cross-checked against the X-ray MicroCT images on which the dense aggregates were initially detected [4,5].

A KB mirror system was used to perform the XFCT measures. This system was designed to produce an intense X-ray microbeam which enables experiments that demand the detection of trace amounts of transition and heavy elements present in small areas from several types of specimens. A white beam focused on a 12 × 22 μm area by the KB system was used to sample excitation. The intensity of the incident beam was monitored with an ionization chamber placed before the sample. The fluorescence photons were collected using two energy-dispersive silicon drift detectors, placed at 90◦ from the incident beam, while transmitted photons were detected with an ionization counter placed behind the sample along the beam path (Figure 2A).

In both experiments, the sample SMC_01 was fixed to the sample holder using albumin (egg white). This substance provides a suitably steady sample fixation while helping to reduce the mounting substrate signal and mitigates the potential for damage to the specimen when it is removed from the sample holder. Then, the fixed sample was placed into the experimental arrangement on a high-precision goniometer and translation stage, which allowed for rotating as well as translating it perpendicularly to the beam (Figure 2A,B). The distance between the sample and both detectors was set at ~27–28 mm, the collimator-detector size was 2 mm, and a 45 µm Al beam filter was used.

The quality of the final reconstructed XFCT image is a compromise between the measuring time required for an acceptable counting statistic of the XRF peaks and the step size necessary to linearly move and rotate the samples. As shown in

Table 1. Experimental parameters for both experiments *.

E	Z (Horizontal Axis)						Y (Vertical Axis)					ϴ (Rotation Axis)
	i mm	f mm	Δ mm	Step mm	Spot n	Time ms	i mm	f mm	Δ mm	Step mm	Spot n	i °	F °	Step °	Spot n
1	0.0	5.3	5.3	0.05	107	200	0.0	3.0	3.0	0.05	61	0	360	6.0	61
2	−6.3	−2.0	4.3	0.025	173	400	0.5	-	-	-	-	0	360	1.5	241
	−6.3	−2.0	4.3	0.025	173	400	0.6	-	-	-	-	0	360	1.5	241
	−6.3	−2.0	4.3	0.025	173	400	0.7	-	-	-	-	0	360	1.5	241
	−6.3	−2.0	4.3	0.025	173	400	0.725	-	-	-	-	0	360	1.5	241
	−6.3	−1.4	4.9	0.025	197	400	10.0	-	-	-	-	0	360	1.5	241
	−6.3	−1.4	4.9	0.025	197	400	12.0	-	-	-	-	0	360	1.5	241
	−6.5	−1.0	5.5	0.025	221	400	18.0	-	-	-	-	0	360	1.5	241
	−6.5	−1.0	5.5	0.025	221	400	24.0	-	-	-	-	0	360	1.5	241

* Experiment (E), initial scanning position (i), final scanning position (f), and scanning distance (Δ).

, the experimental parameters for the first experiment were set to obtain a volume XFCT image of the distal half of the fruit. The same sector was considered for the second experiment, which was set to scan eight 2D individual 22 µm thick transversal slices, gradually spaced from a predefined initial plane (0) at the distal end of the fruit.

The XFCT raw data were fitted using the PyMCA 5.4.1 software [35] mapping the presence and distribution for Ca, K, Mn, Fe, Ti, Si, S, Cu, Br, Rb, and Zn. The in-house RAFT software [36,37,38,39] was used to reconstruct the XFCT slices from both experiments. The lack of 3D information in the second experiment was compensated for by achieving higher resolution images for each of the reconstructed slices. Hence, the pixel size of the final reconstructed XFCT images from the first and second experiments was 50 µm and 25 µm, respectively. The volumes and slices presented in this work were prepared using the image tools of PyMca software [35] and the FIJI distribution of the image processing software ImageJ [40,41,42].

A second exemplar SMC_11 was analyzed using the Energy Dispersive X-ray Spectroscopy (EDS) method in a Scanning Electron Microscopy (SEM) system Jeol^® model 6460LV at the Laboratory of Thin Films in the Institute of Physics of the University of Sao Paulo. These complementary and control SEM-EDS data allow a general view of the chemical components present in an alternative exemplar with the same provenance and taxonomy than SMC_01 while preserving the specimen SMC_01 from the preparation procedures. The sample SMC_11 was broken longitudinally to expose its inner structures and placed in a vacuum camera Denton Vacuun^®. Then, it was vaporized with gold (99.999%), at a pressure of 5 × 10⁻² Pa, to cover the surface with a 10 nm thick gold layer. The metalized sample was fixed with a piece of polycarbonate duct tape to the sample holder of the SEM system, exposing the broken area for measures. The EDS scanning parameters were set as follows: Data Type: Counts; Mag.: 500; Acc. Voltage: 30 kV.

An X-ray absorption MicroCT reconstructed image of SMC_01 was used to correlate and identify dense bodies and some anatomic structures in XFCT images. This reference microtomography was obtained in a Skyscan 1172 (Bruker) scanner at the Laboratory of Applied X-ray Analysis at the State University of Londrina (LARX-UEL). For the experiment, the sample was placed in a PVC tube (straw segment) and immobilized with cotton wool; the PVC tube was fixed to the sample-holder using silicone dough. The main acquisition image parameters were set as follows: Source voltage 33 kV; Source current 198 µA; Exposure time/projection 400 ms, Number of projections 800, Total scan angle 180°; Rotation step 0.250°; Pixel size 1.72 µm; Source to sample distance (FOD) 40.030 mm; Source to detector distance (FDD) 209.887 mm. The software Skyscan NRecon was implemented in the reconstruction process, resulting in a total of 2582 slices with a resolution of 1.72 µm per pixel. The software CTAnalyzer (Bruker) was used for volume visualization and reslicing.

Compositional data from SMC_01 were compared with Energy Dispersive X-ray Fluorescence (ED-XRF) and X-ray Diffraction (XRD) spectroscopy methods applied to five samples of sediment extracted from the stratigraphic layers A, B, C, D, and the feature Burial 1 of Monte Castelo (MC_A, MC_B, MC_C, MC_D, MC_burial). The ED-XRF measures were performed in an XRF system Ampetk composed of a (Ag) X-ray silver tube and an SDD silicon (Si) detector at the Laboratory of Archaeometry and Applied Sciences for Cultural Heritage (LACAPC) of the Institute of Physics in the University of São Paulo (IF-USP). Measures were performed directly on samples with energy 30 kV, current of 5 µA, and exposure time of 200 s. Data fitting was performed using a Windows version of the software system QXAS (WinQXAS 1.4), developed by the International Atomic Energy Agency (IAEA) [43]. Since the experimental setup used has maximum detection efficiency for Fe, the Buffalo River soil reference values [44] were applied to calculate the mass fraction values (%) for some chemical elements of major interest. In the case of P, which is not quantified among the Buffalo River soil reference data, its proportional values in relation to Ca were calculated for each sample.

Lastly, a diffractometer D8 Advance (Bruker) with a K-alpha radiation beam of Cu (λ = 1.54051 Å) was used for XRD measurements of the sediment samples MC_A, MC_B, MC_C, MC_D, and MC_burial at the Laboratory of Crystallography, IF-USP. The experimental parameters were set with energy 40 kV and current 30 mA, at room temperature, with an integration time of 1.5 s in rotative 20 rpm, angular pass 0.02°. The diffractograms were obtained with the 2θ range of 5–75° and were interpreted using the software Diffrac Suite EVA by Bruker and the dataset JCPDS-ICCD version 2003.

4. Results

4.1. ED-XRF

Table 2. ED-XRF results for sediments samples. Areas measured in 200 s.

Z	Energy (keV)	MC_A	MC_B	MC_C	MC_D	MC_Burial
		Area	Area	Area	Area	Area
Si	1.74	1359	1439	1643	675	2034
P	2.013	3128	3060	2909	4884	1686
K	3.313	2386	2252	2688	1825	2777
Ca	3.691	87,240	117,016	100,185	130,791	77,147
Ti	4.509	1558	1157	1754	828	2055
Mn	5.895	3661	2399	2883	1785	2869
Fe	6.399	24,040	28,298	27,857	14,177	28,492
Cu	8.041	477	348	506	439	487
Zn	8.631	2848	2003	2280	2108	2379
Sr	14.142	1169	1356	1455	1760	1331
Zr	15.746	456	471	928	320	580

and Figure 3A show the range of chemical elements present in the analyzed samples and their relative variation through the Bacabal stratigraphic layers of the site. The calculated amount of Ca (29.96%) is the highest in MC_D (Figure 3B) and the P/Ca ratio values suggest a similar pattern for P (calculated on the Area values in Table 2). At the same time, the MC_D sample shows the lowest amounts of Fe (0.67%), K (0.56%), Mn (0.09%), and Ti (0.10%) (Figure 3C). The inverse distribution is observed for the other samples, where lower levels of Ca (17.67%–26.80%) and P are compensated with higher percentages of Fe (1.14%–1.35%), K (0.69%–0.86%), Mn (0.09%–0.19%), and Ti (0.14%–0.24%). The amount of Zn is quite similar in every sample, but the highest one is in MC_A (0.10%).

4.2. XRD

Diffractograms from XRD measures show a noticeable presence of quartz and hydroxyapatite in every sample, particularly in MC_D sediment. By contrast, calcite and aragonite are scarce in this sample as well as in MC_A, where both are practically lacking. The sediment from the burial (MC_burial) shows the highest contents of calcite and rutile, and is the second in terms of quartz and hydroxyapatite contents (Figure 4).

4.3. SEM-EDS

Preliminary compositional data for this study come from SEM-EDS analysis of the alternative sample SMC_11. The four SEM-EDS measures (here expressed in atomic percentages %) show higher amounts of Si (0.68%–4.10%) and Ca (1.06%–2.42%). Other elements such as Al (0.18%–0.99%), P (0.38%–0.73%) and Fe (0.1%–0.27%), followed by Cu (0.04%–0.48%) were observed. K and Ti are also present in the sample SMC_11 (Figure 5A–D) (Table S1). The EDS scan of a delimited area near the distal end of the specimen reveals the presence of Si in cylindrical bodies and P and Ca in angular bodies and other irregular structures (Figure 5E–H). The spatial distribution of P and Ca shows a correlation between these elements.

4.4. XFCT and X-ray MicroCT

The reconstructed volumes from the elemental distribution are consistent with the general morphology of the drupaceous fruit under study (see Figure 1D). The endocarp area concentrates Cu, Br, and Rb, as well as probably Zn, which affects the central cavity and seed/embryo tissues. The mesocarp layer contains a dispersed presence of several elements, but Fe, Mn, and Ti appear more concentrated in 3D reconstructions (Figure 6).

The reconstructed slices from the second XFCT experiment were correlated and compared with X-ray MicroCT imaging data of the same sample. The XFCT transmission images revealed the presence of the major structures already observed in X-ray absorption microtomography (Figure 7). In general, the exocarp and endocarp tissues appear less clearly in the XFCT slices 0.5 mm, 0.6 mm, and 0.7 mm, while their visibility increases gradually toward the last 2.4 mm slice. A total of 8 solid bodies were individualized in both imaging methods and consecutively numbered. Some of the bodies in X-ray MicroCT images were not distinguished in XFCT slices and vice versa. No dense aggregates were observed in 1.8 mm and 2.4 mm slices. The compositional characteristics of the selected dense bodies vary from one to another, as exposed in

Table 3. XFCT results for elemental solid bodies composition.

Dense Structure	Ca	K	Si	Mn	Fe	Ti	Zn	Cu	Br
1	x	x	x	x	x	x	x	x	x
2				x	x				x
3		x		x	x	x	x	x	x
4				x			x	x	x
5	x			x	x		x	x	x
6								x	x
7	x	x	x	x			x	x	x
8							x	x	x
Total	3	3	2	6	4	2	6	7	8

. The most frequent elements in these dense bodies are Br and Cu, followed by Zn, Mn, and Fe.

As shown by the XFCT fluorescence images in Figure 8, Ca, K, Si, and Mn cover the outermost portion of the slices but are absent in the endocarp region. These elements are not limited to the border but also extend over a considerable area inside the mesocarp. By contrast, Cu and Br markedly delineate the endocarp layer, being especially visible for the slices near the middle area of the fruit (1.0 mm, 1.8 mm, and 2.4 mm). Only Zn is clearly present in the central cavity delimited by the innermost layer, over the seed/embryo tissues, as shown in the slice 2.4 mm.

5. Discussion

Results obtained from the ED-XRF analysis of the sediments from the strata A, B, C, D, and the Burial (Phase Bacabal) are mostly consistent with those previously obtained by Pugliese [1], showing some slight increase in the percentage amounts of some minority elements as Fe and K. Accordingly, these data support the prevalence of a characteristic anthropogenic Amazonian Dark Soil (ADS) chemical composition, with large amounts of Ca (17.67% to 29.96%), combined with considerable quantities of Fe (0.67% to 1.35%), K (0.56% to 0.86%), Mn (0.09% to 0.19%), Ti (0.10% to 0.24%), and Zn (0.07% to 0.10%). The burial context from where the analyzed sample came from (sample MC_burial) showed the highest contents of Fe (1.35%) and K (0.86%) and the lowest in Ca (17.67%) and P, inversely to the contiguous Layer D (sample MC_D), where the highest levels of Ca (29.96%) and P combines with decreased quantities of Fe (0.67%) and K (0.56%) and others species like Mn (0.09%), Ti (0.10%) and Zn (0.07%). The XRF spectra and percentage values for the uppermost layers A, B, and C exhibit more similar patterns to those from the burial sediments.

The XRD analysis of the crystalline phase of the soil matrix showed the presence of minerals hydroxyapatite, calcite and aragonite that are consistent with the increased contents of Ca and the presence of P referred above. These characteristics are in line with the abundance of bone material and shell debris at Monte Castelo. The presence of hydroxyapatite crystals in every stratigraphic layer also supports the evidence of Ca and P in solid angular bodies observed by SEM-EDS analysis of the alternative sample SMC_11.

On the other hand, Si and K in ED-XRF spectra refer to the presence of silica and kaolinite minerals observed in XRD analysis, which could have a partly organic origin in the sponge spicules described in previous studies [1]. Additionally, the EDS results from the sample SMC_11 show the occurrence of cylindrical bodies with Si, which could be morphologically assimilable to these spicules fragments (see Figure 5E). In turn, Ti can be related to the presence of rutile crystals. Other elements evidenced in ED-XRF data are absent in XRD results suggesting that Mn, Fe, Zn, and Cu are not present as crystalline compounds but in the form of ions, or in solid solutions with other components, or even, as in the case of Mn, as the product of secondary transformations [9].

Based on this assemblage of chemical elements in the substrate, the compositional and distributional information provided by XFCT slices of the fruit remains reveals a first distinction between the chemical elements associated with the outermost tissues (exocarp and some regions of the mesocarp) and the innermost tissues (seed/embryo, endocarp, and some regions of the mesocarp) in the studied exemplar. On the other hand, solid bodies placed in the middle mesocarp area, in general, share a heterogeneous and more diversified composition.

The outermost region of the drupe, closest to the sedimentary matrix and more exposed to a significant rate of interactions with it, is characterized mostly by the presence of Ca, K, Si, and Mn. Their presence extends inside the sample in some sectors of the mesocarp and endocarp contours. In the case of Ca, it is represented in the XFCT images towards both ends of the endocarp tissue, probably related to the presence of constitutive calcium oxalate biomineralization, already observed by microscopy and X-ray absorption microtomography in several uncharred specimens from Monte Castelo [4,5]. On the other hand, all these elements are notably abundant in sediments, especially Ca, suggesting that at least some of its presence in the fruit could penetrate it through the most exposed tissues, the lax cellular arrangement of the mesocarp and some broken areas.

In turn, the seed/embryo, endocarp, and some mesocarp regions are less exposed structures where exogenous materials are more difficult to access. The elements clearly present there are Rb, Br, Cu, and Zn. The only element associated with the seed/embryo structure is Zn, mostly observable in the slice 2.4 mm, where the structure is bigger. All these elements are also present in areas of the outermost regions of the sample, specifically on the central XFCT slices (1.8 mm and 2.4 mm). However, Rb and Br are only relatable to the plant structure, as they were not found in sediment composition. By contrast, Cu and Zn are present in sediments. For this reason, there is no clear classification of the sedimentary or constitutive provenance for Cu and Zn in the fruit, although they are certainly linked to important plant physiological processes in general [45,46,47,48,49]. In this line, a dual origin is proposed for Cu and Zn, while Rb and Br link better to the plant biology and the soil’s characteristics where these organisms grow, which are not necessarily the same as those of the sedimentary substrate where the fruits remained buried.

Most chemical elements shown by XFCT images are, to a varying degree, present in the dense bodies and alternatively associated with the most external or internal tissues of the plant sample (except Rb, which has not been visualized in the slices experiment). These results suggest that solid aggregates were formed by a combination of exogenous sedimentary material, represented by the chemical elements in ED-XRF results (Ca, K, Si, Mn, Ti, Fe, Cu, and Zn), and those which could be recognized as constitutive of the fruit (especially Rb and Br, but also Cu and Zn). In the case of Fe and Ti, these elements appear primarily in association with solid bodies, being almost absent in other structures. The outermost areas towards the last distal slice present Ti and, to a lesser extent Fe, in a similar pattern to Cu and Zn. Their exogenous provenance is supported by their noticeable abundance in sediments, denoted by the ED-XRF results, especially in those from the burial’s filling material, where the sample was recovered. Finally, SEM-EDS data revealed the presence of P along with Ca in some bigger solid aggregates at the alternative sample SMC_11, as well as some cylindrical bodies formed by Si. Additionally, the SEM-EDS mapping distribution of P and Ca shows a spatial correlation between these elements.

As proposed for the exocarp and adjacent areas of the mesocarp, in contact and closer to the sedimentary matrix, exogenous material could get into the fruit as ions and/or chemical compounds in water solution, or directly as solid bodies, accessing the tissues via the broken areas of the fruit tissues. On the other hand, the lack of Rb and Br in the ED-XRF results of the sedimentary samples, while both are present in the composition of most of the solid bodies and the endocarp, points to an endogenous provenance of these elements. A similar dynamic could have also affected constitutive Cu and Zn from the endocarp and seed/embryo structures. Although a detailed understanding of this characteristic exceeds our data, some studies on experimental taphonomy address the liberation of ions and other chemical compounds as a result of the post-mortem tissues degradation [50]. Decomposition, autolysis, abiotic oxidation, the mobilization of their products, and other mechanisms have been pointed out as effects of the decay and maturation of organisms after their death [9]. These can occur on a continuous or intermittent pattern along with mineralization and interact with it as part of the whole fossilization process [9,50].

6. Conclusions

In general, non-charred archaeological fruits are expected to be composed mainly of C, O, H, and N, followed by small quantities of elements such as K, P, Ca, Mg, S, and Cl, and trace elements that vary in terms of its specific biology and the environment where the plant grows. Once they became part of the archaeological record, as occurred in Monte Castelo, they were exposed to different physicochemical interactions involving their own compositional characteristics and those of the sediments and other materials present at the site.

These interactions occurred during the burial time and contributed to creating a particular range and distribution of different components, especially minerals and trace elements, whose endogenous and exogenous provenance can be traced with varying degrees of certainty. X-ray MicroCT data morphologically revealed the presence of solid bodies which were not relatable to the anatomic structure of the fruit. Their compositional attributes, observed by XFCT provided information about the effective contribution of different materials to their formation. Additionally, it shows the presence of these chemical elements differentially distributed along with the anatomical structures of the fruits.

The XFCT analytical images (and complementary SEM-EDS information) collated and interpreted regarding compositional ED-XRF and XRD data from the sedimentary matrix, suggest that the most externally exposed anatomic structures become enriched in Ca, K, Mn, and Si from the sedimentary substrate. High concentrations of these elements were measured in the sediment samples and identified as hydroxyapatite, calcite, aragonite, kaolinite, and silica compounds.

It was also observed that the distribution pattern of these components within the archaeobotanical sample shows a continuous progression from the exocarp toward the more internal structures. This attribute is associable with some mechanisms of attachment and diffusion or solution impregnation by which exogenous Ca, K, Mn, and Si enter and stay into the sample tissues [51]. However, some of them could also have an endogenous origin as they compose biomineralized substances produced by the fruit itself (i.e., calcium oxalate crystals present in the endocarp structure) [4,5].

The innermost anatomical structures also contain high amounts of Rb and Br, which is not relatable to the sedimentary matrix, based on compositional ED-XRF and XRD data, suggesting their exclusive endogenous provenance. The occurrence of some authigenic precipitation process [52], allowing Rb and Br to reach the more external layers of the fruit, could be proposed, as they are present there in exiguous quantities. Large amounts of Zn and Cu also prevail in the inner tissues, but some exogenous contribution should be considered for these substances, as they were found in sediment composition.

The solid bodies initially observed inside the fruit remains revealed a combined composition that integrates the whole set of elements distributed, along with the anatomical structures and a large fraction of the minerals and trace elements identified in the sedimentary substrate (except Sr and Zr). This supports the idea that dense structures observed in microtomographic images were formed by both sedimentary and endogenous materials. The assembly of chemical elements varies according to different solid bodies, the most frequent being Br, Cu, and Zn, all of them mostly associated with an endogenous provenance, followed by Mn, Fe, and K.

In addition, none of them were relatable to the crystalline compounds identified by XRD in the Monte Castelo matrix, as they have probably been present in the form of ions or solutions mobilized by some mechanical or physicochemical process, such as diffusion or impregnation, and then deposited in the form of concretions inside the fruit. By contrast, the abundance of crystal compounds of Ca, K, Si, and Ti observed in XRD spectra of sedimentary material reveals a lesser significant contribution to dense aggregate formation in SMC_01, although it is effectively present, as evidenced in EDS results on cylindrical and rectangular bodies composed by Si, P and Ca in SMC_11. Analyses also indicate that the crystalline phase of the substrate is more closely associated with minerals and trace elements distributed within the more external tissues of the fruit, which are subject to contact interactions with the sedimentary substrate.

These conclusions reflect a first approach to the alterations that occurred in a non-carbonized plant that persisted for one or two centuries in an open-air site, whose local environment and deposit formation attributes can promote mechanisms toward a gradual and specific acquisition and organization of chemical elements and compounds. These interaction processes between the uncharred plants remain archaeological context and environment suppose a dynamic-in-equilibrium by which their result does not necessarily lead to a successful mineralized preserved material, especially over time spans longer than those considered in this study. Even so, comparable phenomena might lead to different preservation results on other kinds of organic artifacts and materials. Likewise, damage and contamination have also been reported as other possible consequences of mineral precipitation over surfaces and internal structures of organic and inorganic remains [14,53,54]. Tracking the occurrence of these microscale interactions between archaeological artifacts and their environment would provide an opportunity to amplify the understanding about the taphonomy of archaeological organic artifacts and contribute to the site formation process studies.