Phytolith analyses from Khil and Kaf Taht el-Ghar (Western Maghreb): Plant use trajectories in a long-term perspective

Journal of Archaeological Science: Reports 37 (2021) 102921 Available online 1 April 2021 2352-409X/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Phytolith analyses from Khil and Kaf Taht el-Ghar (Western Maghreb): Plant use trajectories in a long-term perspective Carla Lancelotti a , b , c , * , Rafael M. Martínez S´ anchez d , Juan Carlos Vera Rodríguez e , Guillem P´ erez-Jord` a f , Leonor Pe˜ na-Chocarro g , Stefano Biagetti a , c , h , Marco Madella a , b , c , h a CaSEs – Culture and Socio-Ecological Dynamics, Department of Humanities, Universitat Pompeu Fabra, Barcelona, Spain b ICREA, Passeig Lluís Companys 23, Barcelona, Spain c Instituci´ on Mil` a y Fontanals, Departamento de Arqueología y Antropología (CSIC), Barcelona, Spain d Dept. Historia, Facultad de Filosofía y Letras, Universidad de C´ ordoba. 14071 C´ ordoba, Spain e Dept. Historia, Geografía y Antropología. Universidad de Huelva, Spain f GRAMGIUV2015-222. Dept. de Prehistoria, Arqueologia i Hª Antiga. Universitat de Val` encia, Spain g Laboratorio Arqueobiología, Instituto de Historia (CSIC), Madrid, Spain h Department of Geography, Archaeology, and Environmental Studies, University of the Witwatersrand, Johannesburg, South Africa ARTICLE INFO Keywords: Phytoliths Neolithic Crop-processing Palm leaves ABSTRACT In this paper we present the results of phytolith investigations at two archaeological sites in northwestern Morocco: Khil (Tangier) and Kaf Taht el-Ghar (T´ etouan). The two sites located in Western Maghreb, one on the Atlantic and one on the Mediterranean coast, were investigated in the framework of the AGRIWESTMED project. Phytolith analysis complemented archaeobotanical, geoarchaeological and archaeological investigations to better identify plant use during the entire occupation sequence. At Khil 16 samples have been studied, coming from two profiles excavated in two different caves of the same system – grotte B and grotte C – that span a chronology comprised between the early Neolithic and the historic period. At Kaf Taht el-Ghar, 15 samples were analysed spanning from the Palaeolithic to the Historic period. Results show that at Khil plants are widely used from the very beginning of the Early Neolithic sequence. The surrounding vegetation was exploited and both C3 and C4 grasses were used, inside the cave. At the same time a wide array of other plants was utilised, such as palms and woody taxa. Phytolith analysis at the site of Kaf Taht el-Ghar indicates that crop processing was most probably not taking place on site. However, the predominance in the samples of C3 grasses morphologies agrees with the finding of previous macro botanical studies that identified cultivated wheat and barley in this cave deposits. A very interesting aspect of the phytolith assemblages of Kaf Taht el-Ghar is the widespread use of palm leaves and their gradual disappearance through time that coincides with an increase in the use of woody species. 1. Introduction The northern coast of Africa represents the natural interface between the African continent and the Mediterranean Sea. Stretching from the Atlantic shores of Morocco up to Suez, Mediterranean Africa has been a permeable crossroad between Europe, Asia, and Africa from ancient to present times. Due to its geographical location, western Maghreb – the westernmost section of Mediterranean Africa - holds a prominent position in the deep history of contacts and exchanges between Africa and Europe, well before the historical and modern ages, when the Mediterranean Sea became the crossroads of empires, states, and civilizations (e. g. Broodbank, 2013). Recent studies have highlighted the role of the Maghreb in early prehistory (Sahnouni et al., 2018) and the spread of anatomical modern humans (Hublin et al., 2017). Yet, the overall picture of the transition to food production in the Maghreb is still patchy. In this perspective, fresh research has specifically focused on the Holocene in the Maghreb, investigating the arrival of the first Neolithic communities with cereals and legumes towards the middle of the 6th millennium BC (Ballouche and Marinval, 2003; Morales et al., 2013, 2016; Zapata et al., 2013, Martínez-S´ anchez et al., 2018a, 2021). The Tingitana peninsula (where Khil and Kaf Taht el-Ghar are located) is probably one of the best studied areas, with several sites investigated for * Corresponding author. E-mail addresses: [email protected] (C. Lancelotti), [email protected] (R.M. Martínez S´ anchez). Contents lists available at ScienceDirect Journal of Archaeological Science: Reports journal homepage: www.elsevier.com/locate/jasrep https://doi.org/10.1016/j.jasrep.2021.102921 Received 2 October 2020; Received in revised form 24 February 2021; Accepted 5 March 2021 Journal of Archaeological Science: Reports 37 (2021) 102921 2 archaeobotanical remains. A recent overview of the state of research on the process of Neolitisation in this part of North Africa, suggests that domesticated crops were introduced from the southern Iberian Peninsula, together with Cardial wares (Martínez-S´ anchez et al., 2018b). In this paper we present the results of phytolith analysis and we highlight the trajectories of plant exploitation and use at the sites of Khil and Kaf That el-Ghar, from the Paleolithic to Historic times. 1.1. Context, environment and chronology of the samples analysed The two archaeological sites under study are located in the Tingitana peninsula, the northernmost region of Morocco. Khil is placed on the North Atlantic coast and Kaf Taht el-Ghar next to the Mediterranean shore (Fig. 1). At both sites the prevalent climate is Mediterranean with oceanic influences, with the highest rainfall between October and March and frequent dry spells during summer. The Aridity Index (Tabucco and Zomer, 2019; Zomer et al., 2007, 2008) at both sites is c. 0.55 defining the environment as dry sub-humid. Thus, today the prevalent vegetation at both sites is the maquis-forest (Morales et al., 2016 and references therein). Nowadays, North Africa is much drier than it was in the Early and Middle Holocene, when the environment was more humid and ‘green’. Different proxies point to climatic oscillations that determined changes in both sea-level and vegetation, variably affecting ancient human societies (DeMenocal, 2001; Kuper and Kr¨ opelin, 2006; Clarke et al., 2016). Sea level varied considerably from the Last Glacial Maximum (c. 23-20kya), when the sea was up to 130 m lower than by 4000–3000 BCE, when it reached its current level (e.g, Benjamin et al., 2017). Shifts in sea-level reflect more general climatic trends that affected North Africa throughout the Holocene. Yet, along with regional environmental dynamics, local phenomena must have played an important role and need to be considered as well, since the physiography of specific setting (e.g., mountains, oases) might have shaped local cultural dynamics (e.g. Cremaschi et al., 2014). A recent macro-scale approach has reviewed the available data from the different regions of North Africa and pinpointed three main phases marked by arid spells recorded continentally to better fit social, environmental, and economic information available: Phase 1 - from the beginning of the Holocene up to the 6200 cal. BC climatic arid spell, characterized by warm climate and by a foraging economy based also on marine resources; Phase 2 – from 6200 to c. 4000 cal. BC, when the spread of domesticates from the Near East rapidly reached the Maghreb following the onset of improved environmental conditions; finally Phase 3 – from 4000 to c. 1000 cal. BC, which still remain poorly understood in the Maghreb due to the lack of archaeological stratigraphies, and when the environment turned to present conditions. (Broodbank and Lucarini, 2019, and references therein). Morocco, especially its eastern portion, has been the focus of several archaeobotanical investigations in recent years. Such investigations have identified the long persistence of a mixed hunter-gatherers-fishers economy even during the Neolithic with no definite presence of domestic plants and animals, contrary to what happens in western Maghreb (Portillo et al., 2020 and references therein). The extensive archaeobotanical analyses of charred macro-remains carried out at the sites of Khil and Kaf Taht el-Ghar (Morales et al., 2016) show that cereals (emmer, free-threshing wheat, naked barley), pulses (broad bean) and fruits (grape, mastic tree and myrtle) were being consumed at these sites by 5500–5000 cal. BC. Similarly, at the site of Ifri n’Amr ou Moussa, wild pulses and fruit of the mastic tree were identified in the Epipaleolithic layers, whereas in the Neolithic phase (dated to 5100 cal. BC) seeds of domesticated cereals are found together with seeds and fruit of wild Fig. 1. Location of the two archaeological sites anlaysed in the present work based on a map of Aridity Index values (data extracted from Zomer et al. 2007, 2008). C. Lancelotti et al. Journal of Archaeological Science: Reports 37 (2021) 102921 3 arboreal species (Carri´ on Marco et al., 2018). 1.1.1. Khil The archaeological site of Khil is composed by a series of caves situated at the end of the Ashakar Wadi, in Tangier province (Figs. 1 and 2). It is a karstic formation where several caves of small dimensions were inhabited, or sporadically used, in the past by small communities of agro-pastoralists, fishers and shell-gatherers. From the archaeological evidence it seems possible to place the beginning of the frequentation of this site in the mid-VI millennium, with a peak during the fifth millennium, and occupation continuing up to historical times (Pe˜ na-Chocarro et al., 2012; Martínez-S´ anchez et al., 2018b). In September 2011 a joined team of Spanish and Moroccan archaeologists revisited the site in the framework of a collaborative project between the Spanish National Research Council (CSIC) and the National Institute for the Archaeological and Cultural Heritage Sciences (INSAP). The project aimed at acquiring data to evaluate the role of the first agricultural communities in the area through the systematic analysis of bioarchaeological material (charred seeds and wood, pollen and phytoliths, faunal and micro-faunal remains) collected during the new excavations. On this occasion interventions were carried out in three caves -grotte B, C and Dand systematic samples were collected from the two that showed cultural material corresponding to the Neolithic period, grotte B and C, in line with the aims of the study. The macro-botanical remains are dominated by broad bean (Vicia faba) and naked wheat (Triticum aestivum/durum) grains (Morales et al., 2016). Radiocarbon dates on short-lived specimens confirm an Early Neolithic phase dated to c. 5300–5000 cal. BC and a Middle Neolithic phase sated to c. 4500–4000 cal. BC (Martínez-S´ anchez et al., 2018a; Morales et al., 2016). The proposed chronology of the two caves, with correlations between the two sequences can be found in Table 1. Sixteen samples have been studied, coming from the two profiles (Fig. 2 and Table 1). Four samples from grotte C belong to the historic period and were analysed as control samples. Of the remaining samples six were dated to the Middle Neolithic period (three from grotte C and three from grotte B) and six to the Early Neolithic (five from grotte C and one from grotte B). All samples were collected from the general deposit of the cave and represent different moments of anthropic occupation. 1.1.2. Kaf Taht el-Ghar The archaeological site of Kaf Taht el-Ghar is located in the province of T´ etouan (Figs. 1 and 2). It is a karstic cave, situated in the B´ eniHosmar massif dominating over the Oued Martil Valley and the Mediterranean coast (Ballouche and Marinval, 2003). The site was studied first in the fifties by a Spanish archaeological mission (Tarradell, 195758) and then in the late eighties by a French-Moroccan archaeological team (Daugas et al., 2008) that established it as one of the oldest Neolithic sites in Northern Africa. Traces of domesticated species, both of plants and animals, were identified in the Early Neolithic levels (Ballouche and Marinval, 2003). The authors found domesticated emmer, einkorn and naked wheat (Triticum dicoccum, T. monococcum and T. aestivum/durum) in a layer dated to the end of the VI millennium Cal BC. The AGRIWESTMED team revisited the site in 2012 (MartínezS´ anchez et al., 2018a, 2021) when four profiles were cleaned and sampled for bioarchaeological samples. In this occasion up to 42 stratigraphic units were identified and bulk sediment samples were collected for phytolith analysis from three of the four profiles (26 AB, 26 G and 26 Fig. 2. Stratigraphy of the two sites under study and location of samples analysed. C. Lancelotti et al. Journal of Archaeological Science: Reports 37 (2021) 102921 4 JK, see Martínez-S´ anchez et al., 2021 for the full description of the excavation material and methods and details on the excavated trenches). For the present work, 16 samples were processed of which only 15 were then used for analysis (Fig. 2 and Table 2). One sample was excluded because of the few phytoliths observed, which all belonged to categories that are not taxonomically significant. All the fifteen samples analysed come from the anthropogenic sediment of the cave, from layers of occupational deposition. Two of them were collected from fireplaces and were studied separately as they represent different types of deposits that involve specific strategies of plant use. The other samples examined cover the entire chronological sequence from the Palaeolithic to Historic periods. 2. Material and methods A total of 31 samples were analysed: 16 from Khil and 15 from Kaf Taht el-Ghar covering the entire chronology of occupation at both sites (see Tables 1, 2 for samples description and Tables 3 and 4 for details on phytoliths). Samples were collected from recently excavated sections at both sites. Phytoliths were extracted from bulk sediment samples following the procedure described by Madella et al., (1998) slightly modified to calculate the AIF (Acid-Insoluble Fraction, Albert and Weiner, 2001) and to adapt to recent studies about the aggressive effect of too concentrated chemicals and long times of exposure (Cabanes et al., 2011). The sediment was dried in a drying cabinet until no loss of weight was recorded, then 4 to 5 g of material were subsampled for analysis. The full protocol used for extraction as well as raw data can be found in the Supplementary Material (SM1, SM2 and SM3) and are stored on GitHub. Phytolith were mounted in permanent medium and observed under transmitted light microscope with magnification between 20X and 40X. Phytoliths were counted up to 300 identified individual morphotypes or, when not possible, the full slide was scanned. Identification was based on published reference material, such as thePhytCore database (2021) Table 1 Periods and phases of the two caves at El Khil, their stratigraphic correlation as defined during excavation (from Pe˜ na-Chocarro et al. 2012, Martínez-S´ anchez et al 2018a) and description of the contexts sampled. In bold the stratigraphic units analysed for this work. Radiocarbon dates from Martínez-S´ anchez et al. 2018a and Morales et al. 2016. Period Phase grotte B grotte C 1 Sterile layers with evidence of bioturbation (s.u. 1, 2) Sub-actual 2 Post Medieval, scarce evidence of occupation; high bioturbation (s.u. 3) Historic 3 Islamic Medieval occupation, marine resource exploitation (s.u. 4, 5) 4 Scarce or no human presence (s.u. 6–8) 5 Late Roman occupation with ceramic dated to the IVV centuries AD (s.u. 9) Abandonment 1 6 Transition between prehistoric and historic periods (s.u. 10) Abandonment Late prehistory 2 7 Protohistoric occupation (s.u 2001, 2002) Recent prehistory, sporadic use, possible funerary use (s.u. 11) LatePrehistory Neolithic 3 8 Post-depositional erosion processes (s.u. 2003). Intense occupation with Ashakar ware and abundant marine resources (s.u. 2004, 2005, 2006, 2007, 2008)5490 ±40 (Beta 316505); 5420 ±40 (Beta 316506) Frequent occupation with characteristic Neolithic material (both lithic and ceramic) and remains of domesticated crops and animals (s.u. 12,13,14)5450 ±40 (Beta 316507); 5470 ±40 (Beta 316508) MiddleNeolithic Early Neolithic B 4 9 Beginning of the occupation of grotte B, with non Cardial impressions and with evident signs of marine resources exploitation(s.u. 2009, 2010)5790 ±30 (Beta 331845) EarlyNeolithic Early Neolithic A 4 10 Cardial impressions and channelled ware (s.u. 15, 16, 17, 18) 6180 ±50 BP (Beta-295780) Natural layer Alteration of the natural rock by biological agents and weathering. S.u. 19 contains large quantity of fossil malacofauna and lays on sterile sand u.s 20 Geologic Table 2 Periods and phases of the excavated trenches at Kef That el Ghar, their stratigraphic correlation and dating as defined during excavation, and description of the contexts sampled (from Martínez-S´ anchez et al., 2021 and references therein). In bold the stratigraphic units analysed for this work. Phase 26AB / 26AB ext 26G 26HI / 26JK Historic 1 Long phase that contains Medieval wheel-made pottery alongside some later prehistory potsherds. s.u. 1100, 1100 hearth, 1101–1, 1101–2 s.u.1201 Middle Neolithic 2 The most recent of the prehistoric sequences, shows a less intense occupation and is dated c. 4200 cal. BC. Identified plant remains include naked wheat (Triticum aestivum/durum), legumes (Lathyrus/Vicia) and fruits (Pistacia lentiscus and Vitis vinifera). 1101–3, 1102 1201–1, 1202–2, 1203 Early Neolithic 3 The richest phase in terms of material culture and plant and animal remains, dated between 5500 and 5100 cal. BC. Identified plant remains include cereals (Triticum diccocum, T. aestivum/durom, Hordeum vulgare), legumes (Lathyrus/Vicia sspp.) and fruits (Vitis vinifera and Myrtus communis). 1103–1, 1103–2 1003–1 1204–1, 1204–2 Transition 4 Period of abandonment or very low frequentation of the cave, dated between 6200 and 5500 cal. BC. s.u. 1104 s.u.1003 Crust s.u.1205 Epipaleolithic 5 This phase consists of repeated occupations presenting at least 4 hearths, one floor level and a possible post hole and is dated between 10,900 and 6200 cal. BC. It yielded abundant microlithic industry, but no pottery and no domestic plant or animal remains. s.u.1104 concretion, 1105 s.u. 1003–2, 1003 Hearth, 1003 Under Hearth, 1003–3, 1003–4, 1003–5, 1003–6, 1003–7, 1004, 1005, 1019, 1006–1, 1006–2, 1006 Hearth, 1006 Pit, 1006–3, 1006 Ash, 1006–4, 1007 Interfacies, 1006 Basis s.u. 1208, 1209–1, 1209–2, 1209–3 Paleolithic 6 Long phase that spans from the Middle Pleistocene to the Upper Palaeolithic subdivided into 2 periods (s.u. 1024, 1025). Evident frequentation with bone remains and lithic artefacts C. Lancelotti et al. Journal of Archaeological Science: Reports 37 (2021) 102921 5 and naming follows ICPN 1.0 (Madella et al., 2005) 3. Results 3.1. Phytolith taphonomy 3.1.1. Khil The phytolith assemblages from Khil generally presents a low level of taphonomic damage. Indeed, the number of elongated cells identified is quite high (Fig. 3) thus advocating for a generally good preservation rate (Madella and Lancelotti, 2012). Some evident signs of pitting, typical of chemical erosion, and mechanical stress have been observed on the phytoliths during analysis, both on elongated as well as on more resistant morphotypes like bulliforms and trichomes (Fig. 4). Taphonomic processes seem to have had a stronger effect on the assemblages of the later periods in samples from grotte C, while no evident sign of damage has been observed on phytoliths from grotte B or the earlier periods of grotte C. Average values of phytolith indicators are generally higher for grotte B than grotte C: phytolith concentration (463 k/195 k); number of morphotypes (15/7); inflorescence morphotypes (43/11); leaf/culm morphotypes (81/8.5); woody morphotypes (9.75/3.25); palm morphotypes (73.5/3.8). The relationship between the concentration of phytolith per gram of AIF and number of morphotypes identified in the assemblage shows a low degree of correlation (Fig. 5). This indicates that post-depositional taphonomic processes do not affect the richness of the phytolith assemblages in a significant way (Madella and Lancelotti, 2012). 3.1.2. Kaf Taht el -Ghar The phytolith assemblages observed in Kaf Taht el Ghar show a higher rate of damage due to post-depositional processes. As summarised in Fig. 6, short cells are definitely predominant in the assemblages, while elongated forms are rather scarce and silica skeletons almost absent. However, the assemblages show quite a high number of morphotypes and there is no direct correlation between the phytolith concentration and the number of morphotypes identified in the samples (Fig. 7). Silica skeletons are present mostly in the Epipaleolithic samples, both in the general sediments and in the two fireplaces. 3.2. Phytolith concentration and morphology 3.2.1. Khil The assemblages at Khil are characterised by a marked difference between grotte B and grotte C both in terms of phytolith concentration and morphology. The sediments collected in grotte B are much richer in phytolith content as well as in number of morphotypes identified (Table 3). Indeed, grotte B shows an average number of morphotypes that doubles that of grotte C. There is a generalized increase in plant diversity starting from the end of the Early Neolithic period and through Table 3 Khil: summary of the samples analysed and the results of the analysis. Concentration Morphotypes n. Inflorescence Leaf/culm Woody taxa Palms Period Sample Value mean Value mean Value mean Value mean Value mean Value mean Post Neolithic EKH-C-8 819,799 568,508 14 11 42 31.5 28 17 5 6.5 3 6.75 EKH-C-9 1,174,836 9 57 18 4 14 EKH-C-10 275,039 11 26 6 8 8 EKH-C-11 4,359 10 1 16 9 2 Middle Neolithic EKH-C-12 542 115,463 2 9 0 10.6 1 40.6 0 6.6 0 38.6 EKH-C-13 1,561 2 0 4 2 0 EKH-C-13/2 2,101 4 0 3 7 0 EKH-B-2005 322,565 15 6 100 18 165 EKH-C-14 37,050 8 0 3 2 5 EKH-B-2006 201,985 17 22 112 9 22 EKH-B-2008 242,436 15 46 61 8 78 Early Neolithic B EKH-B-2009 1,084,783 13 97 51 4 0 29 29 Early Neolithic A EKH-C-15/2–3 16,881 7,215 7 6.75 0 0.75 9 0.75 0 0.5 5 3.5 EKH-C-15/3 2,624 8 2 8 0 3 EKH-C-17 1,661 6 0 4 2 0 EKH-C-18 7,694 6 1 1 0 6 Table 4 Kaf Taht el-Ghar: summary of the samples analysed and the results of the analysis. The two fireplace samples are dated to the Epipalaeolithic period but have been kept separate from the others as they represent a different type of deposit and their formation correspond to different depositional processes than those of the other samples. Concentration Morphotypes n. Inflorescence Leaf/culm Woody taxa Palms Period Sample Value mean Value mean Value mean Value mean Value mean Value mean Historic KTG 1.12 2,416,331 11 7 36 12 13 Middle Neolithic KTG 1.10 589,063 713,566 11 11 3 2 11 10 4 13.5 30 21 KTG 1.8 838,068 11 1 9 23 12 Early Neolithic B KTG 1.6 505,049 482,145 8 10.7 0 1 10 11.7 37 21.7 15 14 KTG 1.3 445,668 12 1 9 20 1 KTG 2.9 495,179 12 2 16 8 26 Early Neolithic A KTG 2.7 33,483 421,636 8 6.3 0 0 6 6 12 5 27 40 KTG 2.5 59,128 5 0 2 2 3 KTG 3.16 1,172,298 6 0 10 1 90 Transition KTG 2.2 946,340 4 0 12 0 178 Epipalaeolithic KTG 3.12 1,659,625 7,034,134 9 9 0 1 11 11.5 4 5.5 364 297.5 KTG 3.10 12,408,643 9 2 12 7 231 Palaeolithic KTG 3.4 5,386 5,451 2 1.5 0 0 0 0 0 0 0 2 KTG 3.2 5,561 1 0 0 0 4 Fireplaces(Epipaleolithic) KTG 3.x 7,690,415 14,283,901 11 10 0 0 12 10 8 5 269 288 KTG 3.xx 20,877,386 9 0 8 2 307 C. Lancelotti et al. Journal of Archaeological Science: Reports 37 (2021) 102921 6 all the Middle Neolithic Period (Fig. 8). This increase is mainly due to the assemblages of grotte B, whereas in grotte C there seems to be very little activity related to plants until much later, during the Roman frequentation of the cave. Sample EKH-B-2009, from grotte B, is particularly noticeable as it shows an overall concentration that is as high as the deposits of historic period of grotte C and it presents an amount of inflorescence indicators (elongate echinates and dendritics) that is highest among the samples studied. In addition, the silica skeletons of inflorescence are clearly of two different types indicating the presence of both C3 (wheat and barley type) and C4 (millet type) species. The presence of both C3 and C4 plants at the site is reflected also in the short cells assemblages (Fig. 9). The subfamily Pooideae is by far the most represented however, chloridoids and panicoids are present, especially in the assemblages of grotte B. It is interesting to note the presence of exclusively panicoid type short cells in the assemblage of EKH-C-18, although the absolute amount is rather low, and the evidence is not statistically significant due to the scarcity of phytoliths observed in this sample (n =14). No panicoid short cells were observed in sample EKH-B-2009. However, in this same sample, silica skeletons of C4 inflorescence were observed thus indicating that this group of plants was collected since the earlier occupation of grotte B. Silica skeletons were in general scarce in the assemblages studied (Fig. 3), with the notable exceptions of samples EKH-C-15/3 and EKH-B2009. Both contain silica skeletons of grasses’ inflorescence and leaf/ culms, much more abundant in the sample coming from grotte B than in the one coming from grotte C. 3.2.2. Kaf Taht el-Ghar Phytolith concentration per gram of AIF is in general quite low in the sediments of Kaf Taht el-Ghar (Table 4). The two samples from fireplaces have double the concentration of the Epipalaeolithic general samples, which present the highest concentration among the occupation levels. If fireplaces are considered, only these and the Epipalaeolithic samples show a concentration above the mean. When the mean is calculated excluding the fireplaces, only two occupational moments show a concentration above the mean: Epipalaeolithic and Historic times. This is not per se an indication of taphonomy, but it suggests that plants were not commonly used inside the cave. A total of 19 different morphotypes have been observed in the samples from Kaf That el Gar. In general, and excluding the two fireplaces, the number of morphotypes tends to increase in the most recent samples. Indeed, from a mean of 1.5 in the Palaeolithic levels the number increases to 11 in the Middle Neolithic and Historic levels. The notable exceptions are again the Epipalaeolithic samples: the two fireplaces with a mean of 10 and the general occupations with 9 morphotypes (Table 4). The most abundant morphotype observed by far in all sample is the globular echinate produced by the leaves of Arecaceae family (palms - Fig. 10b-d-e-g-h-). In Fig. 11 these are represented on a scale 100 times higher than for all other morphotypes and are particularly abundant in the Epipalaeolithic fireplaces and in the earlier Neolithic levels. There is seemingly a decrease in the presence of palm phytoliths as time progresses, up to the historic levels where they are very scarce. At the same time,woody indicators (globular psilate, globular granulate, scalloped, sclereids, parallelepipedal and irregular forms), although less represented in absolute terms, seem to follow exactly an opposite trend. Their number starts to increase more or less when palms indicators start to decrease. Fig. 3. Khil: proportion of elongate cells, short cells and silica skeletons identified in the samples. C. Lancelotti et al. Journal of Archaeological Science: Reports 37 (2021) 102921 7 Fig. 4. Khil: microphotographs of significant phytoliths morphotypes. A) globular echinate and trichome; B) Trichome; C) Elongate psilate; D) Elongate echinate; E) Wavy -top rondel; F) Globular echinate; G) Silica skeleton (2 cells) of elongate echinate cells; H) Elongate not determined; I) Elongate (possible echinate). All the phytoliths in this figure (except the globular echinates in A and F and the rondel in E) showing signs of chemical weathering. Scalebar in B =20 µm (valid for all photogrpahs). 5 10 15 20 0 400,000 800,000 1,200,000 concentration number of morphotypes Fig. 5. Khil: correlation between the number of morphotypes identified during analysis (y axis) and the phytolith concentration per gram of AIF (x axis). C. Lancelotti et al. Journal of Archaeological Science: Reports 37 (2021) 102921 8 KTG_3.xx KTG_3.x KTG_3.2 KTG_3.10 KTG_3.12 KTG_2.2 KTG_3.16 KTG_2.5 KTG_2.7 KTG_2.9 KTG_1.3 KTG_1.6 KTG_1.8 KTG_1.10 KTG_1.12 0% 25% 50% 75% 100% sample elongate_cells short_cells silica_skeletons Fig. 6. Kaf Taht el-Ghar: proportion of elongate cells, short cells and silica skeletons identified in the samples. 4 8 12 16 0 5,000,000 10,000,00015,000,00020,000,000 concentration number of morphotypes Fig. 7. Kaf Taht el-Ghar: correlation between the number of morphotypes identified during analysis (y axis) and the phytolith concentration per gram of AIF (x axis). C. Lancelotti et al. Journal of Archaeological Science: Reports 37 (2021) 102921 9 Grasses are scarcely present in all the samples analysed and are in general represented by leaf/culm phytoliths (elongate psilate, elongate sinuate and bulliforms). Inflorescence indicators (elongate echinate and dendritics) are very few and practically only present in the upper levels of the Middle Neolithic and Historic periods. In these same levels, leaf/ culm indicators increase as well, following the same trend as phytolith concentration. Three subfamilies of grasses are represented in the samples, Pooideae, Chloridoideae and Panicoideae (Fig. 12) although Pooideae are predominant. Again, the highest variability is found in the fireplaces and in the Historic period samples. Interesting to note the presence of Panicoideae morphotypes (in this case two bilobates) in sample KTG 1.3 as they are the only encountered in all samples. However, the number of these morphotypes is so small that they are not statistically significant. 4. Discussion 4.1. Evidence of cereal processing and use The results of phytolith analysis complement the information offered by charred macro-remains and add some interesting insights on the use of plant resources, as well as the general environmental trends in the Tingitana peninsula of Morocco during the Holocene. Especially at Khil, the general concentration of phytoliths, and the high number of morphotypes encountered, indicate that plant-related activities were definitely taking place inside grotte B. Morales et al., (2016) reports only a few macrobotanical remains from this deposit, the majority of the evidence at Khil being represented by two concentrations of broad beans recovered from grotte C, possibly the result of one episode of accidental burning. The concentration and diversity of phytoliths recovered from grotte B indicate, in accordance with the presence and abundance of cultural material recovered during excavation (Martínez-S´ anchez et al., 2018a), that this cave was occupied more intensely than grotte C; in addition, it suggests that plant-related activities took place preferentially in this location. The high amount of both inflorescence and leaf/ culm morphotypes shows that entire plants were brought into the cave. It is possible that plants were then processed on site to separate the leaves and culms from the edible parts but both components seem to have been used intensively at this location. Leaf/and culms could have been employed as bedding material or fodder for the animals while people consumed the grains. It cannot be excluded that phytoliths could have also entered the archaeological record via the deposition of animal dung. Rests of animal dung have been identified in contemporaneous archaeological sites in the region (Carri´ on Marco et al., 2018). However, the combined results of macrobotanical and microbotanical analyses, together with the characteristics of the phytolith assemblages, suggests that the observed phytoliths mostly arise from an important use of plants at the site. The few silica skeletons encountered in the sediments of grotte B advocates for the exploitation of two different groups of grasses were exploited during the Neolithic period at El Khil: C3 types, such as wheat/ barley and C4 types, such as millets (a group that includes wild and cultivated types). Studies have been published with guidelines to identify C3 cerals (e.g., Ball et al., 1999; Portillo et al., 2006) and millets (e. Fig. 8. Khil: chronological variation of the main categories of phytoliths observed in the samples from grotte B and grotte C. Note that the scales are different as woody taxa produce fewer phytoliths than the other categories. C. Lancelotti et al.