Multistep food plant processing at Grotta Paglicci (Southern Italy) around 32,600 cal B.P.

Grotta Paglicci [Rignano Garganico, Apulia, N 41.6539722, E 15.6151666, 143 m above sea level (a.s.l.)] is located on the eastern slope of the Vallone di Settepenne, a valley excavated in the karst formation overlooking the Candelaro River plain, at the foot of the Gargano promontory, a mountainous peninsula jutting into the Adriatic sea on the southeastern coast of Italy, characterized by a Mediterranean climate (Supporting Information and Fig. S1).

Excavations at Paglicci have been carried out for over 40 y, first by the Museo Civico di Storia Naturale di Verona and, since 1971, by the University of Siena, in collaboration with the Soprintendenza Archeologia della Puglia. The stratigraphic sequence of the cave (12 m) (Fig. S2) yielded a large quantity of Paleolithic artifacts, examples of portable art, burials, and human and animal remains. Wall paintings of horses and hands were discovered in the innermost part of the cave. These parietal paintings are the only Paleolithic ones known in Italy (8). The retrieved materials mostly date to the Upper Paleolithic, namely to the Aurignacian, with marginally backed bladelets, and to the entire Gravettian and Epigravettian chronology (9, 10).

The faunal association, along with the results of the isotopic analyses from faunal remains, indicate a climate colder than the present one (Supporting Information).

Plant remains consist of charcoal fragments mostly belonging to Juniperus, Pistacia, and Prunus, occurring in almost all of the Aurignacian and the Gravettian levels. Deciduous Quercus was often recorded. Pinus spp., Salix/Populus, Acer, Rhamnus, and Fraxinus sporadically occurred (11).

The tool (Fig. 1) was found in June 1989 in sublayer 23A (square 36H) (Supporting Information and Fig. S2). This sublayer was radiocarbon dated 28,100 ± 400 B.P. [UtC-1414: 32,614 ± 429 calibrated (cal) B.P.] (12, 13).

On the basis of the functional analysis, this item has been interpreted as a pestle-grinder; evidence of the grinding function was found mainly on side 2 (13). Since its retrieval, this artifact has been stored in a plastic bag. It is an elongated cobble of fine, not completely cemented calcareous sandstone bearing longitudinal fissures, fractured at one end, slender and rounded at the other end, with a flat subquadrangular section. It measures 11.8 × 5.5 × 3.7 cm and weighs 443 g.

The stone, not previously washed, was subjected to two successive samplings (Fig. 1). The first one was carried out in 2010 on the apical area [sample P23A(a)] and on the medial band area [P23A(b)]. In 2012, following a new procedure, three further samples were collected: P23A(c) on side 2, close to the apex; P23A(d) from the previously washed medial band area on side 2; and P23A(e) from an unwashed, presumably unused area on side 1, near to the fractured end of the stone.

Residue analyses revealed the presence of numerous starch grains concentrated on the apex [samples P23A(a)], close to the apex [P23A(c)], and on the medial band of the tool [sample P23A(b)]. More specifically, 250 starch grains per cm² were calculated in sample P23A(c). In sample P23A(d), which was collected from side 2 after washing, and in sample P23A(e), from an unused area on side 1, 25 grains per cm² and 40 grains per cm² were, respectively, detected.

Most of the grains (over 60%) showed more or less pronounced fissures radiating from the center of the grains, possibly referable to the grinding process; numerous grains revealed a faded or very faded extinction cross with a very wide center. Gelatinized, swollen grains (about 4% of the total grains) were observed in samples P23A(a), P23A(b), and P23A(c) (Supporting Information, Fig. S3, and Table S1).

Excluding the swollen grains, five different starch morphotypes were recorded, in addition to the rather numerous spherical–ovoid small grains (≤5 μm in size), which were not taken into consideration in view of the lack of any diagnostic feature for identification and because they are very common in plants. The morphotypes proved to be unevenly distributed in the examined samples (Fig. 2).

Residue analysis revealed the presence of a large number of grains on the Paglicci tool, providing new information on the food plants that were selected and processed and the procedures used for producing flour.

This research demonstrates a specific skill in processing plant foods during the Early Gravettian, casting light on complex human behavior.

The sampling method, which was adopted here for the first time, to our knowledge, on a pestle-grinder, has furnished additional elements for the functional study of these kinds of artifacts, in terms of more precise indications about the quantity and dimensions of the starch grains in relation to the use traces.

The production of flour requires multistep processing and manipulation before cooking, depending on the different parts of the plants used. Grinding requires a previous drying of the plant portion to be processed, and drying may be accelerated by means of heat treatment. The present analysis indicates that the inhabitants of Grotta Paglicci (sublayer 23A) were the most ancient population to use a method that involves at least four subsequent steps in preparing plants for consumption. The examination performed on the Paglicci pestle-grinder provides direct evidence of heating and grinding. Although there is no direct evidence of the following steps, namely the mixing of the flour with water and the cooking, these processes can be plausibly hypothesized because the rehydration is necessary for cooking and the cooking is necessary to make the starch digestible. The thermal pretreatment could have been an additional crucial step in a period that was characterized by a climate colder than the current one.

The majority of the starch grains recovered on the Paglicci pestle-grinder indicate a preference for Poaceae. The collection from the wild of grass caryopses, as opposed to underground organs, and the subsequent processing phases of heating, grinding, and preparing for cooking were undoubtedly time-consuming activities. These observations demonstrate that, from at least the Early Gravettian, economic activities connected with the exploitation of vegetable resources for nutritional purposes had assumed an important role in the subsistence strategies of these peoples.

It is significant that the plant most represented on the pestle-grinder is an Avena species. This finding is currently the most ancient evidence of the processing of Avena, a genus that includes A. sativa, a cereal that was to be a successful crop long after. In the absence of spontaneous forms of other cereals that were later domesticated in the Near East and that feature a better ratio of time investment/energy yield, oats could have represented an interesting nutritional source for the diet of European Gravettian groups.

In conclusion, this research clearly reveals that the early modern humans of Paglicci already had a wealth of well-developed knowledge that included a multistep technology and organizational ability in processing food plants, long before the Neolithic.

In 2010, sampling for residue analysis was performed according to Revedin et al. (3), using a light jet of distilled water directed on two areas (a and b) of the cobble, selected based on the occurrence of use-wear traces (Fig. 1). The sampling on the apex (a) was performed by precisely directing the jet of water at the apex of the tool: the dashed line in Fig. 1 indicates the unwrapped area that became wet even though not directly struck by the jet of water. Sample b involved a central band of the artifact, where both sides were directly struck by the jet of water. In 2012, further samplings (c, d, and e) were carried out according to a procedure (Fig. 1) previously tested on experimental grinding tools: the stone was wrapped in a PVC plastic film, leaving a selected square area 2 × 2 cm free for collecting samples. This procedure was refined with the aim of obtaining a quantitative evaluation of the residues and their distribution on the tool’s surface, to acquire information regarding the motion of the artifact during grinding. Moreover, the wrapping preserves a large part of the grinding stone surface for further analyses. The number of plant residues per cm² was calculated in 30 μl of each sample (1 ml). The three areas sampled using the new method were chosen in relation to the position and type of the use traces: sample c, located on side 2 close to the apex, already involved in the previous sampling (a) in proximity with the traces of percussion; sample d, located in the center of side 2, at the point corresponding to the use traces (striation) and the previous sampling (b); and sample e, located on side 1, in proximity with the base of the tool, in an area without any use traces.

Samples P23A(a) and P23A(b) were then subjected to heavy liquid separation, using zinc chloride, and mounted in a water–glycerol 50% (vol/vol) solution. Samples P23A(c)–P23A(e) were directly mounted in water. Analyses were carried out under a light microscope, as is, or after iodine/potassium iodide (Lugol) staining (37), and a polarizing microscope, operating at ×400 magnification. Overall, about 1,000 starch grains were observed. All of the grains were measured; mean value and mode are reported for the morphotypes exceeding the number 50. Identification was performed with the aid of literature (19, 23) and modern reference collections of the Department of Biology in Florence and the Herbarium Centrale Italicum. The modern material was selected on the basis of the available data on the ancient vegetation at Paglicci (11) and the plants currently common in comparable environmental contexts (Fagus, Castanea, Quercus spp., Pistacia terebinthus, Festuca spp., Stipa spp., Ammophila arenaria, Dasypyrum villosum, Rumex acetosa, Glyceria maxima, Phragmites australis, Nymphaea, Nuphar, and Typha).

The Gargano peninsula (Fig. S1) is at present dominated by the Mediterranean macroclimate, more or less markedly influenced by the complex topography. The highest part of the Gargano is characterized by a subcontinental climate, with the mean temperatures of the coldest months (January and February) ranging between 8 °C and 11 °C (38). Below 300–400 m a.s.l., the climate is milder in the areas influenced by the proximity to the sea whereas, in the inland areas, where the Paglicci site is located, the winters remain cold, as in the mountainous regions of the Gargano.

The potential vegetation can be referred to as Quercus virgiliana woods (39). This type of vegetation prefers deep soil such as “Terra Rossa.” Together with Q. virgiliana, other deciduous trees can be found: Fraxinus ornus, Carpinus orientalis, and Ostrya carpinifolia. Deciduous and evergreen Mediterranean sclerophyllous plants form the shrub layer. On the colluvial soils at the bottom of the valleys, Q. virgiliana woods lack the most thermoxerophylous species, and the deciduous components increase in biomass. In both cases, the vegetation of substitution is represented by shrublands dominated by Paliurus spina-christi, Cercis siliquaster, Prunus spinosa, Crataegus monogyna and Pyrus amygdaliformis. Grasslands can be referred to the Festuco-Brometea association with steppa species: e.g., Stipa sp. Along the permanent rivers, Populus alba, Fraxinus oxycarpa, and/or Salix sp.pl. may also occur.

Level 23 (Early Gravettian) (Fig. S2), with thickness reaching a maximum of 70 cm, was excavated by A. Palma di Cesnola between 1989 and 1993 over 20 m² and was divided into different sublayers (A, B, and C, the latter partially divided into C1 and C2), all characterized by yellowish-brown sandy-silty sediments, mixed with angular stones and collapsed blocks (40, 41).

In sublayer 23 A (<30 cm), evidence for human occupation is clearly documented by the presence of a pit filled with charcoal, ashes, and burned stones from a hearth, a few bone/antler points, pigments, perforated animal teeth and marine shells, a single human tooth (a deciduous lower left incisor), and a right temporal bone (42). The large quantity of lithic material has already been studied from a typological (40), techno-functional (43), and technological viewpoint (44).

In the Paglicci stratigraphic sequence, stable isotope studies on bones and teeth revealed that major cooling events took place at about 29,000 and 26,500 cal B.P. and at about 23,000 and 19,500 cal B.P., with repeated minor oscillation at different levels; a shift from arid to humid conditions probably occurred between 23,500 and 19,000 cal B.P. (45, 46).

Micro and macro animal remains from the lower part of the Upper Paleolithic (Aurignacian levels) indicate the presence of a dry-temperate phase with a forest steppe that evolved into a xeric cold steppe at about 34,000 cal B.P. The presence of ungulates (mainly Bos primigenius and Capra ibex) and equids (Equus ferus and Equus hydruntinus) also suggests the local occurrence of prairies and steppes during the Early Gravettian period. In sublayer 23A, the elevated presence of Bos primigenius, the decrease of C. ibex and the increase of Microtus savii testify a milder phase (47–49).

With the aim of reproducing the morphological changes observed in the Paglicci starch grains, modern caryopses of Panicum miliaceum and Avena sativa were subjected to thermal treatment: popping and roasting on the surface of a stone placed over an open fire for 5 min, and for a longer time until the caryopses changed color (13 min P. miliaceum; 15 min A. sativa). After these treatments, the caryopses were ground (this process proved much easier than before the thermal treatment), and the starch grains were observed under a light microscope (Table S1). Some of the starch grains proved to be gelatinized and swollen; in Panicum, their amount ranged from 5% to 30%, depending on the length and temperature of the treatment, but, after a period (about 3 mo) of permanence in water/glycerol solution, about 50% of the grains were swollen (Fig. S3). The complete gelatinization of the grains in the caryopsis may have been prevented by the presence of a coat (lemma and palea) that inhibits efficient heat transfer, as observed after the popping experiment on millet starch grains (50).

Three Avena species are native in the North Mediterranean: the tetraploid species A. barbata Pott ex Link, sectio Aristulatae Malz., and the hexaploid species A. fatua L. and A. sterilis L., both belonging to the sectio Denticulatae Malz. Avena barbata is distributed throughout the Middle East and all of the Mediterranean region; Avena fatua is considered to be a Euro-Siberian-Western and Central Asian element, which is widespread in the Mediterranean region; Avena sterilis is an East Mediterranean-Irano-Turanian element, common in the warm part of the Mediterranean Basin, extending up to the Atlantic coasts (14, 15). Indeed, the hexaploid oat species are fully interfertile, the hybrids naturally occurring where the geographical distribution areas overlap. Therefore, Ladizinsky (51) has recently proposed to regard all of the hexaploid oats as a single species.

Avena has spherical to ovoid compound starch grains, which are formed by numerous polyhedral individual grains or seldom two-three-few tightly clustered grains, and simple spherical–subspherical grains (Fig. S4). In A. barbata, the polyhedral grains present the main axis ranging from 5 to 20 μm; in A. fatua and A. sterilis, most of them do not exceed 5 μm (Table S2).

A remarkable variety of starch grains were observed in the Italian species of Quercus (Fig. S5): spherical to subspherical grains, 2.5–15 μm in diameter, with centric hilum; ovoid grains 2–20 μm in length (up to 30 μm in Quercus trojana), with slightly eccentric hilum; tear-shaped grains, 5–23 μm in length, with eccentric hilum; bell-shaped grains, 2.5–15 μm in length, with centric hilum; rare reniform grains, 5–18 μm in length (up to 25 μm in Quercus ilex); and elongated, irregularly ovoid grains with irregular outline, 5–30 μm in length. All of the morphotypes may display point-shaped, linear, or X,Y-shaped hilum; a more or less evident longitudinal fissure may be present in the elongated grains. The relative number of morphotypes differs in the examined species.

A preliminary screening carried out on oak acorns (Table S3) showed that the dimensional range of the irregular grains differs in the examined species. The largest irregular grains belong to deciduous Quercus. At Paglicci, the presence of deciduous Quercus was recorded by means of charcoal analysis (11).

In addition to the starch grains, a few other plant remains were found in the samples (Fig. S6): rare phytoliths, mostly fragmented: spheroidal verrucate, bulliform fan-shaped from stems or leaves of grasses, elongate psilate; isolated fibers, Angiosperm vessels, and Gymnosperm tracheids with bordered pits; diatoms belonging to Pennales and Centrales; and pollen grains in a very bad state of preservation. Among the latter, grains belonging to Pinaceae, deciduous Quercus, and Chenopodiaceae were observed.

We thank Dr. G. Giachi (Soprintendenza Archeologia della Toscana), Prof. A. Brandi (Dipartimento di Chimica “Ugo Schiff,” University of Florence), P. Boscato, and A. Moroni (University of Siena) for critical reading and constructive comments; Mrs. T. Gonnelli for technical assistance; and the Museum of Natural History of Florence, Herbarium Centrale Italicum for reference materials. We also thank the two anonymous reviewers for suggestions that contributed to the improvement of the paper. We thank the Soprintendenza Archeologia della Puglia, which authorized and supported the excavations at Grotta Paglicci, and Prof. A. Palma di Cesnola for fieldwork and studies over the years. The research was carried out with the support of the Istituto Italiano di Preistoria e Protostoria, the Ente Cassa di Risparmio di Firenze, and the University of Florence (Fondi di Ateneo per la ricerca scientifica, ex 60%). This research is part of the “Vegetable Resources in the Palaeolithic” project.