1. Introduction

1In Klimonas, building earth samples from the PPNA Communal building (St 10; Mylona et al. 2017a, chap. 11) and palaoeosoils interbedded either in the fluvial terraces, located a few hundred metres south of the site in the sedimentary sequences, or the “road profile”, at the eastern edge of the site (Mylona et al. 2017b, chap. 3), were selected for phytolith analysis.

2Phytoliths are microscopic silica opal particles that form in the tissues during the plant’s life. Because of the physical and chemical properties of the opal, they are well preserved in most sedimentary contexts (Lebreton et al. 2017). Grass (Poaceae) produce large quantities of phytoliths and various different morphotypes, allowing not only taxonomic identifications but also the recognition of plant parts (straw, leaves, husk; Carnelli et al. 2001). The so-called “grass silica short-cell phytoliths” (GSSCP) exhibit variable morphologies according to the grass subfamilies (Festucoideae, Panicoideae, Chloridoideae; Twiss et al. 1969, ICPT 2019). When fragments of silicified epidermal tissue are preserved in the form of multi-cell phytoliths, also known as silica skeletons, the identification of cereals can reach the genus or even the species level (e.g. Rosen 1992, Emery-Barbier and Thiébault 2005, Ryan 2011, Madella et al. 2014).

3Earthen building material may contain phytoliths from the raw earth or from plants added intentionally, in order to modify its properties (temper), or unintentionally during earth material preparation or drying (plant fragments trapped inside/on the earth). Phytolith analysis of the building earth from Klimonas aimed to document the composition of this material and to provide data on the use of plant resources in a context where organic remains are poorly preserved (chap. 23). The study of pedo-sedimentary sequences from the colluvial terraces aimed to record the local vegetation around the site and its evolution through time. The existence of several palaeosoils gave access to potential areas of plant gathering, or even real agricultural activities, for contemporary populations.

4All the building earth samples proved to be rich in well-preserved phytoliths. On the contrary, samples from the natural sequences were extremely poor, even when pedogenized levels were involved (appendix 22-1).

2. Method

5In order to extract phytoliths, the fine fraction (200µm) of the sediment was treated first with concentrated hydrochloric acid and then with potassium hydroxide in an ultrasonic hot bath. After several rinses, the phytoliths were concentrated by heavy liquid separation using a Thoulet solution (density = 2.35). They were then mounted on a microscope slide in a drop of immersion oil and identification was carried out through a transmitted light microscope, at a magnification of 1000x. Where needed, polarizer/analyser devices were used to differentiate amorphous silica opal from bi-refringent particles.

6This microscopic observation allowed the ascribing of phytoliths to previously described morphotypes whose taxonomic and anatomic significances are already documented (ICPT 2019). More than 250 identified phytoliths were counted for each sample, except when they were not abundant enough. In these cases, that only concerned the natural sequences, the data is purely qualitative.

3. Phytolith analysis of building earth

3.1. Phytolith spectra

7Nineteen samples of building earth were studied (appendix 22-1, fig. 22-1). Most of the identified morphotypes were produced by grasses, but some particles that characterize dicotyledons and palms were also encountered. As no palm species is reported to be native from Cyprus, it is likely that the presence of the palm echinate morphotype attests to a low -contribution of airborne phytoliths in the spectra.

Fig. 22-1 – Phytolith diagram of building earth samples and sediment from the natural sequences. Brown: phytoliths from grass inflorescence bracts, yellow: phytoliths from grass vegetative parts, green: dicots and palms.

8The phytolith assemblages show little variation from one level to another. The only noticeable difference is the relative amounts of short (GSSCP) and long cells among grass phytoliths. Multi-celled phytoliths appeared in moderate but constant proportions in all the samples (0.9–6.1%). When phytoliths are naturally present in the earth used to prepare building material, they consist as individual particles, released during the decomposition of plant tissues. The constant and significant (although moderate) presence of articulated silicified cells in Klimonas building earth is consistent with fresh plant material added to the building earth during processing. The very low abundance of phytoliths in the samples from natural sequences (infra) suggests that the raw earth processed was in all likelihood poor in plant silica particles, supporting the hypothesis that the phytolith spectra from the earthen material is almost entirely attributable to the input of vegetal tempering material (see also chap. 11 and 23). Consequently, the phytolith spectra testifies more to human practices than to local vegetation.

9Grass silica short cells were abundant. They refer mainly to two subfamillies: the Festucoideae, which mainly produces trapeziform, conical and conical-polylobate short cells; and the Panicoideae, which mainly produces bilobate short cells. The so-called “saddle” morphotype, typical of the Chloridoideae subfamily, was much rarer. Grasses from the Festucoideae subfamily exhibit a C₃ metabolism and thrive in temperate areas. Most of the cereals originated from the Near East (wheat, barley, oats) and nearly all the wild European grasses belong to the Festucoideae subfamily. Panicoid grasses are adapted to warmer conditions and mostly exhibit a C₄ metabolic pattern. They are mainly distributed over intertropical areas, except for a few species, such as wild and domestic millets (Setaria sp., Panicum sp.) that spread over the Mediterranean or even the temperate areas, where they nevertheless only account for a very small part of the herbaceous flora. In Cyprus, besides millets, the genus Digitaria, Echinochloa and Andropogon can be found. Finally, the Chloridoideae subfamily is clearly of tropical affinity and is therefore weakly represented in the Mediterranean. In Cyprus, the genus Crypsis, Sporobolus, Cynodon, Aeluropus are reported (http://www.flora-of-cyprus.eu/).

10Elongate morphotypes account for an important part of the phytolith assemblages (29–55 %), mainly in the form of elongate entire phytoliths. These are produced in large quantities in grass stems, and to a lesser extent in leaves, and help to keep the plant erect. Elongate dentate morphotypes occur in the leaves epidermis. In the inflorescences, a continuum can be observed from the dentate to typical dendritic ornamentations, the latter being exclusive to ear bracts. The dendritic and dentate forms were particularly abundant in the building earth of Klimonas (10–20 %; mean 15.6 %, of which 11.2 % dendritic and 4.4 % dentate). Such amounts evidence that husks significantly contributed to the phytolith record. It has been shown that when grass straw (either cereal or wild species) is gathered without any contribution of husks, either because only stems and leaves are selected or because the plant is collected before flowering, typical husk phytoliths only account for one to a few percent of the assemblage (Lancelotti et al. 2014, Delhon et al. 2019). The abundance of dendritic long cells (close to 10%) suggests that inflorescence bracts contributed to the phytolith assemblage, while proportions over 30 % were considered as an evidence of cereal processing, either wild (Emery-Barbier and Thiébault 2005) or cultivated (Cabanes et al. 2009). According to Novello and Barboni (2015), the fact that dendritic morphotypes largely overpass 3 % of the sum of grass silica short cells and dendritic elongate is a signature of anthropogenic activity. In Klimonas building material, this value is on average around 18 % (ranging from 7.6 to 43.1 %).

11Thus, the phytolith spectra from the Klimonas building material attests to the abundance of husks in the vegetal temper, which could point to the use of grain by-product processing, like chaff. A closer observation of multi-celled phytoliths gives further arguments.

3.2. Silica skeletons: cereal identification and break mark profiles

12Sixteen samples of building earth from the Communal building (St 10) were selected for specific silica skeleton analysis, in order to identify the species – especially if cereals were involved – that could have been used as temper in the building earth, and to search for any possible stigmata left by the mechanical processing of the grasses. The anatomical origin (stem/leaf or husk) of around 100 multi-celled phytoliths for each sample was determined. Whenever possible, the characteristics of the observed silicified cells were compared to published descriptions of cereals phytoliths (Rosen 1992; Ball et al. 1996, 2001). Finally, the break mark profiles of each silica skeleton was described (shape and orientation; fig. 22-2).

Fig. 22-2 – Silica skeletons from the building earth.

13In all the samples, multi-cellular fragments of the silicified epidermis presenting dendritic cells whose shape corresponded to that of wheat, barley (Triticum sp., Hordeum sp.: fig. 22-2, A, B) and non-cereal wild grasses were observed. Noticeably, the presence of Phragmites (reed) and Cyperaceae phytoliths evidenced the use of wetland plants (fig. 22-2, E, F). Some morphologies evoked that of rye (Secale sp., fig. 22-2, C), which is not attested among the plant macroremains at Klimonas (chap. 23) nor on any site of the island predating the Pottery Neolithic (Lucas and Fuller 2020). However, on the continent, the PPNA site of Jerf el-Ahmar (Syria) revealed the remains of wild rye (Willcox and Fornite 1999; Willcox and Stordeur 2012), making its occurrence at Klimonas plausible. In any event, the phytolith data should be consolidated to ensure the identification of that genus, which has been less extensively described than Triticum and Hordeum in phytolith literature.

14Husk silica skeletons were noticeably abundant and large-sized in samples from US 10.2, St 118, St 144 and St 166. In the latter, they were often composed of 10 to 20 long cells, sometimes more than 30. The size of silica skeletons, given by the number of articulated cells that compose them, largely relies on the level of tissue silification. Hence, it is linked to maturity of the ears (silica impregnation of cell walls increases with time), genetic factors and, above all, on environmental conditions, as a good supply of water favours the optimal silicification of glume cells (Rosen and Weiner 1994). Furthermore, preservation of the large silica skeletons produced by the plant, once it has been torn to chaff, also depends on the aggressiveness of the processing methods, particularly during dehusking or grinding (Portillo et al. 2013, 2017). Mechanical actions such as cutting or crushing the organic matrix before its decomposition generate typical break mark profiles on multi-celled phytoliths which are different from those resulting from natural decay (Anderson 2003, Avner et al. 2003, Cummings 2003, Anderson et al. 2006, Danu et al. 2019). The silica skeletons extracted from the building earth of Klimonas, however, displayed mainly straight–smooth or concave profiles whose trajectories were not directed by cell boundaries (fig. 22-2, G, H). Stepped profiles, following some of the cell walls, were less frequent (fig. 22-3). Straight–smooth and concave profiles are only produced by mechanical treatments, linked to dehusking or the grinding of grains. Break mark profiles, together with the taxonomic attribution of some of the phytoliths, show that at least part of the vegetal material used as temper in the Klimonas building earth was made from the by-products of wild or cultivated cereal processing. The macroremains and the impressions left by temper in the cob also attest to the use of cereal chaff (chap. 11 and 23). This result echoes the phytolith study at the PPNB site of Çatalhöyük (Central Anatolia), suggesting that “the site was made of plants as well as clay” (Ryan 2011).

Fig. 22-3 – Proportions of break mark profiles on the silica skeletons.

4. Phytolith analysis of sedimentary sequences

15The sediment from the alluvial and colluvial sequences, located in the fluvial terraces or at the edge of the Klimonas site, turned to be almost sterile in phytoliths (tab. 22-1). Knowing that they preserve well in most of the sedimentary contexts, an extreme paucity of phytoliths usually characterizes deposits that formed without any contribution (or with a very low contribution) of plant material. In contrast, palaeosoils usually contain various amounts of phytoliths originating from the plants involved in pedogenesis. Phytolith concentration can be high where dense vegetation composed of silica rich plants (grasses, palms, etc) is concerned. On the contrary, it can remain rather low if the vegetal cover is sparse and/or if the plants involved produced few or no phytoliths (dicotyledons). In the Late Glacial and Early Holocene fluvial terraces close to the site, and in the earliest palaeosoils of the sedimentary “road profile” sequence, which date to the Late Glacial (chap. 3), phytolith concentrations are so low that for most of the samples it has not been possible to obtain quantitative results (Riboud 2017). Data is sufficient only in the uppermost palaeosoil of the “road profile”, which is the only sample that allowed the calculation of relative abundancies (appendix 22-1). Though poor in phytoliths, the preparations contained many radiolarian capsule fragments (siliceous microfossils of marine protozoa) of local geological origin (Moni limestone, chap. 3). Their morphology makes them easily recognizable, but confusion is still possible between small isolated fragments of radiolaria and some globular phytoliths produced by dicotyledons, in particular the cystoliths. Thus, the reported possible presence of dicotyledons phytoliths in the samples should be considered cautiously, especially in the absence of pedogenetic features.

16The only interpretable sample comes from the most recent palaeosoil, which is also the one whose datation is closest to that of the PPNA village of Klimonas. Nevertheless, based on the available ¹⁴C date (8872±37 BP, 8224-7837 BC), it appears to be a few centuries younger than the archaeological occupation. This may be due to a slight contamination of the organic matter by more recent plant roots (chap. 3). Thus it cannot be excluded that this palaeosoil is more or less strictly contemporaneous with the occupation of the Klimonas village. The phytolith spectrum records a vegetation largely dominated by grasses, mainly from the Festucoideae subfamily (rondel, trapezoid and polylobate short cells) and to a lesser extent the Panicoideae subfamily (bilobate short cells). Elongate entire and sinuate grass phytoliths produced in the vegetative parts of the plants (stem and leaves) are unusually scarce. Four multi-celled silica skeletons were observed (including one from a husk epidermis), which is not anecdotal considering the usual scarcity of these morphotypes outside specific archaeological contexts, generally linked with cereal storage or processing. The elongate dendritic morphotype accounts for 10.7 % of the assemblage; this is a high score for natural grassland (meadow, steppe) for which values of a few percent, at the most, are expected (see for example Pető 2013, Dal Corso et al. 2018). Novello and Barboni (2015), who studied surface assemblages from sub-Saharan Africa (Chad), in an area where useful grass species, including wild panicoid cereals, are abundant, observed that dendritic long cells represent less than 3 % of the total phytolith counts from modern non-cultivated soils.

17The accumulation of inflorescence bract phytoliths in the palaeosoil CRPL3 makes the hypothesis for cereals growing on this plot of land. Non-cereal grasses produce less developed surface ornamentation that makes elongate inflorescence bract phytoliths fall into the “elongate dentate” morphotype rather than the “elongate dendritic” one. Moreover, the proportion of total biomass represented by ears is more important in cereals than in most wild grasses, which is one of the reasons why husk phytoliths are proportionally more abundant in that group. The absence of other usable samples in the natural sequences makes it difficult to have a clear-cut interpretation of the palaeosoil’s phytolith spectrum. Nevertheless, the fact that the phytoliths almost exclusively come from grasses, together with the abundance of dendritic forms, make it likely that cereals largely contributed to the phytolith record. It is possible, therefore, that this plot of land hosted a cultivated – or at least exploited – cereal field in the course of the 9^th millennium BC, either during the occupation of the village (cultivations near or even inside the village) or after the abandonment.

5. Conclusion

18The use of herbaceous plants for non-food purposes represents an often underestimated part of the prehistoric plant economy. In Klimonas, the recipe for earthen building material includes an important quantity of plants (chap. 11), partly in the form of by-products of crop processing. Cereals were collected in order to respond to various needs, rather than just for food, and went through various technical systems; however, the data does not permit us to determine to what extent they could have been cultivated (see also chap. 23). These cereals grew at least partly on soils sufficiently watered to insure a high level of silification in epidemic cells. The use of wild plants from damp soils (reeds, sedges) evidences the exploitation of rivers, or lakesides, which is consistent with the presence of freshwater crabs (chap. 24) and waterbirds (chap. 26) in the archaeozoological records of Klimonas. Among the studied palaeosoils, one displays a phytolith spectrum compatible with that expected in a cereal field, but its strict contemporaneity with the short period of occupation of the PPNA village cannot be ascertained.

19The use of chaff as temper in building earth is widespread in populations who grow cereals and have an earthen architecture. In Çatalhöyük (PPNB, Turkey), the use of cereal chaff, together with leaves and stems from wild grasses, is interpreted as the opportunistic exploitation of materials available throughout the year, at least whenever glume cereals stored in the form of spikelets where involved (Ryan 2011). In that respect, the chaff supplied could have notably relied on emmer and einkorn, whose presence is attested by micromorphological analyses (Mylona et al. 2017a; chap. 11) and by macroremains (chap. 23).

20Hence, despite the scarcity of preserved plant macroremains, phytolith analysis points to the importance of cereals in the economy, and to the integration of the chaîne opératoire for the manufacture of building materials into a technical system linked with the (proto)-agricultural system.