Life between
lake and land
Life between lake and land
Local vegetation reconstruction and the
potential for plant exploitation at the Middle
Pleistocene lake-shore site Marathousa-1,
Greece
Oda Nuij Bachelor thesis, 1043SCR1Y-17-18ARCH Supervisor: Dr. M. H. Field Specialisation: Botany Leiden University, Faculty of Archaeology Leiden, 15 June 2018, final version
- 2 -
Index
Acknowledgements 4
1. Introduction 5
1.1 Marathousa-1 6
1.1.1 Geology of the Megalopolis basin and Marathousa-1 7 1.1.2 Palaeontological and archaeological finds 10
1.1.3 Age of the site 16
1.2 Research questions and relevance 17
2. Materials and methods 19
2.1 Sampling in the field 19
2.1.1 Description of the section and samples 19
2.2 Data acquisition in the lab 21
2.3 Data analysis: semi-quantitative or concentration 22
2.4 Plant use: literature review 23
3. Results of the macro-botanical analysis 24
3.1 Semi-quantitative results: general characteristics 24 3.1.1 Some remarks on certain species 28
3.2 Vegetation change? 28
3.2.1 Properties of the sediment 28
3.2.2 Semi-quantitative nature of the dataset 29
3.2.3 Subjective analysis 30
3.3 Sample 1: a comparison between semi-quantitative data and 30 concentration data
3.3.1 Observed differences and possible explanations for 31 them
3.3.2 Information gain vs. time investment 32 3.4 Conclusions on the macro-botanical analysis 33
- 3 -
4. Discussion on Palaeolithic plant exploitation 34 4.1 Sources to reconstruct Palaeolithic plant use 34 4.1.1 Information from non-botanical sources 34 4.1.2 Information from preserved botanical remains 36
4.2 Hominin plant use in the Palaeolithic 38
4.2.1 Food 38
4.2.2 Raw material supply 39
4.2.3 Fire 40
4.3 Possible plant use at Marathousa-1 40
4.3.1 Food 41
4.3.2 Raw material supply 43
4.3.3 Fire 44
4.4 Conclusions on the potential for plant exploitation at 46 Marathousa-1
5. Conclusions 47
5.1 Suggestions for further research 48
Abstract 49
Bibliography 50
List of figures, tables and appendices 57
Figures 57
Tables 58
Appendices 58
Appendices 59
Appendix 1: alphabetical list of all plant taxa found at MAR-1 59
- 4 -
Acknowledgements
This thesis wouldn’t have been here without the following people, whom I’d like to thank.
Mike, for letting me work on this material, for all the help with identifying, for learning me almost everything I know about macrofossils, for the discussions about how to present “whishy-whooshy” subjects like this one, for trusting me that although I didn’t give you a first draft it would all turn out alright, and for the regular horse-talks. Erica, for identifying the wood fragments, for all the fun times in the lab (or, rather, keeping all the time I spent in the lab fun) and for your encouragement.
Everybody else who was working next to me in the lab, struggling to get our samples finished on time. Knowing that I wasn’t the only one made it easier to keep on picking. All students who followed the BA3 Botany specialisation, for all your work on the samples, especially Eleni and Nina for helping with “my” sample 1.
Simone, Lars, Nina and Sarah for their feedback on my lines of thought regarding the site, structuring a thesis, Palaeolithic plant use etc.
The people of the Crossroads project for allowing some Leiden students to sieve and sort all the excavated sediments on site, in Greece (thanks Mike for arranging this). Being at the site and handling the material helped me a great deal in understanding what I was writing about.
And my parents for being interested in what I’m doing, asking how the lab work and writing was going, with my mother helping me with some Dutch plant names and ecologies of plant taxa.
- 5 -
1. Introduction
During the Early Pleistocene, faunal movement between Africa and Europe took place through the Levantine Corridor (Martínez-Navarro 2010, 218). Although many aspects of “Out of Africa I” (the idea that Homo erectus evolved in East Africa and settled in Asia in the Early Pleistocene) are currently debated (Dennell 2010), it is likely that hominins moving out of Africa into Eurasia (and vice versa) also used the Levantine Corridor, since the dispersal of early hominins followed the same pattern as that of other large
mammals (Belmaker 2010, 198). From the Levantine Corridor onwards, the commonly named routes into Europe are across mainland Turkey and the Bosporus Straight or through the Caucasus (Tourloukis and Karkanas 2012, 12). However, considering that much of the Aegean Sea would have been dry land during both glacials and interglacials up to MIS (Marine Isotope Stage) 10 and during glacials from MIS 10 onwards
(Tourloukis and Karkanas 2012, 9), other possible routes into Europe are envisaged, which all go through Greece (Tourloukis and Karkanas 2012, 12). Apart from being at the crossroads between Europe, Asia and Africa, there are two other reasons why Greece holds a prominent position in early hominin dispersals: the Balkan peninsula is believed to have served as an important refugium for flora, fauna and therefore likely also hominins during Pleistocene cold periods; and current evidence indicates that southern Europe was inhabited earlier than the part of Europe north of the Alps (Tourloukis 2010, 11). Sadly, an estimated 95-98% of the Early Pleistocene record of Greece is probably not preserved due to tectonic activity, a markedly seasonal climate, changes in sea-level and various slope processes that occurred in Greece throughout the Pleistocene and are still active in the Holocene (Tourloukis and Karkanas 2012, 13). Of the potentially
preserved 2-5%, much of the material is expected to be found either in primary contexts that are inaccessible because they are buried deeply under younger sediments, or in accessible but reworked contexts (Tourloukis and Karkanas 2012, 13). The most promising places to find Lower Palaeolithic material in a stratified primary context are low-gradient palaeo-surfaces in basins, preferably inverted due to uplift, since these places combine the two prerequisites: preservation and accessibility/visibility (Tourloukis 2010, 208). One such place is the Megalopolis basin in the central
- 6 -
Peloponnesus (fig. 1). In 2012-2013, a targeted survey was carried out here,
investigating paleo-lake systems with the goal of finding evidence for Early and Middle Pleistocene hominin activity in intact stratified contexts (Thompson et al. 2018, 1). The survey was conducted by a joint team of the University of Tübingen and the Ephoreia of Paleoanthropology and Speleology of the Greek ministry of Culture, in the framework of the ERC-funded PaGE (“Paleoanthropology at the Gates of Europe”) project (Tourloukis and Harvati 2017, 2). In a region measuring 14km by 13km, a total area of 1.40km2 was
surveyed, resulting in the discovery of ten sites. Three of these sites yielded lithics, five contained palaeontological finds and two provided a combination of both find
categories (Thompson et al. 2018, 4-5). Of the ten discovered sites, Marathousa-1 is by far the most rich in finds (Thompson et al. 2018, 8).
1.1 Marathousa-1
The site Marathousa-1 (MAR-1) is located in a dormant east facing section profile of the Marathousa lignite mine (37˚24’32”N 22˚05’29”E) (fig. 1). Fossil animal bones and stratigraphically associated lithics were recovered along several hundred meters of the profile. The discovery of these remains led to the rescue (2013) and subsequent systematic excavations (2014-present) of the site (Thompson et al. 2018, 8). Two excavation trenches, approximately 60m apart, were opened: one on top of a scatter of elephant bones eroding out of the profile, named Area A, and another one above the highest density of lithics, named Area B. Excavations at the site will go on till 2019, in the framework of the ERC-funded Crossroads (“Human Evolution at the Crossroads”) project (Harvati 2018 pers. comm.).
- 7 -
Figure 1: the location of the Megalopolis basin and the site MAR-1 (after Panagopoulou et al. 2015, 1).
1.1.1 Geology of the Megalopolis basin and Marathousa-1
The Megalopolis basin is a graben. The bedrock of the basin was formed from the Paleozoic to the Eocene and consists of a mixture of marine sedimentary rocks, ultramafic rocks and crystalline rocks. During the Pliocene the graben itself formed by NNW-SSE trending primary normal faults that were followed by secondary NE-SW faults in the Pleistocene, leading to tectonic subsidence (Karkanas et al. 2018, 2). The basin has a surface height of between 400m and 450m above sea level (masl) and is surrounded by hills of up to 1300masl (Vinken 1965, 101). Due to the tectonic subsidence, a lake formed in the basin, which only dried up at some point probably in the Late Pleistocene when the Alfeios river drained it (Touloukis 2010, 110). From the late Pliocene onwards different layers of sediment were deposited in the graben (table 1; fig. 2a). Of particular
- 8 -
interest here is the Marathousa member of the Choremi formation, in which MAR-1 is situated.
Table 1: the stratigraphy of the sediments in the Megalopolis basin (after Karkanas et al. 2018, 2; Vinken 1965, 109).
Period Formation Member Sedimentary environment Sediment type
Holocene Holocene fluviatile terrace gravel
Pleistocene Thoknia fluviatile terrace gravel
Pleistocene Potamia fluviatile terrace gravel; debris
Pleistocene Choremi Megalopolis fluviatile gravel; debris
Marathousa lacustrine clay; sand; lignite
Pleistocene Apiditsa alluvial fan gravel; debris
Pliocene Trilofon fluviatile gravel; sand/marl
Pliocene Makrision lacustrine marl; lignite
The lacustrine Marathousa member consists of four 10-20m thick lignite seams (labelled I to IV from the bottom up) divided by detrital beds of clay, silt and sand of 5-30m thick (Van Vugt et al. 2000, 80). Palynology shows that the lignite seams are characterized by interglacial southern European flora and that the detrital beds in between represent glacial flora of the Eastern Mediterranean. This indicates that the lignite seams were deposited during interglacials and the detrital beds during glacials (Okuda et al. 2002, 152).
The geological layer in which Marathousa-1 is found, is the detrital bed between lignite seams II and III. The detrital bed consists of several sedimentary units and differs
between Area A and Area B. It is thickest in Area B and becomes thinner towards Area A. The units within the sedimentary sequence are labelled top down, having the letter U for unit, then the letter of the area followed by the number of the unit (fig. 2b). The sequence is divided into two parts, the lower and upper, which are separated by a major erosional contact (Karkanas et al. 2018, 3). The lower part of the sequence in both areas is characterized by a variable depositional environment, but sedimentation was continuous (Karkanas et al. 2018, 3). In Area B (UB7-9), the sediments of the lower part are dominated by grey to bluish silts and sands that were accumulated in the deeper
- 9 -
parts of the lake. UB9 is a massive layer of organic-rich clayey sand. Both UB8 and UB7 are layers of interbedded silt and fine sand and display deformation structures (Karkanas
et al. 2018, 5). In Area A, the lower part of the sequence (units UA5-6) is made up of
dark and bluish grey massive sands which were deposited at the margins of the lake. UA5 contains deformation structures (Karkanas et al. 2018, 5). The top of the sequence’s lower part is formed by a massive layer with load deformation structures that is visible in both areas: unit UB6 and UA4. This boudinaged layer consists of eroded remnants of bluish to grey muddy sand and is best preserved in area B, whereas in area A it is discontinuous. It was deposited in the marsh at the edge of the lake and liquefied and slumped after deposition. It was also affected by waves and currents (Karkanas et al. 2018, 4-5).
The upper part of the sedimentary sequence is different from the lower part. The units in the upper part are separated by erosional contacts and consist of coarser sediments at the base and progressively finer sediments upwards (Karkanas et al. 2018, 4). Unit UB5 is dark grey in colour. The unit is subdivided into three subunits: a, b and c, labelled top-down. UB5c and UB5b are laminated sands and have local intraclast-rich areas. These subunits are the result of fluviatile flows entering the lake margins. Subunit UB5a is massive, rich in organics and interpreted as hyperconcentrated flows which were emplaced under water. In Area A there is no unit comparable to UB5 (Karkanas et al. 2018, 5). The following units UB4a-c and UA3a-c are present in both areas and share the same characteristics. UB4c and UA3c consist of massive dark grey sands, rich in
intraclasts. They are the result of dilute mudflows and hyperconcentrated flows that plunged into the lake margin (Karkanas et al. 2018, 5). UB4b and UA3b are massive organic rich silty sands with occasional sand laminae, resulting from subaqueously emplaced hyperconcentrated flows with intervening discrete surges. Subunits UB4a and UA3a are made up of massive organic rich muddy sands, which emanated from
subaqueously emplaced hyperconcentrated flows (Karkanas et al. 2018, 5). In area B, UB4a is followed by unit UB3, which is a massive dark grey silty sand layer. The last unit of the sedimentary sequence in both areas (UB2 and UA2) is recognised by a layer with high amounts of shell fragments interbedded with silty sands. In Area A, these
sediments are deposited at the margins of the lake, whereas in area B they are partly deposited in somewhat deeper water (Karkanas et al. 2018, 4-5). Most of the
- 10 -
archaeological material (see section 1.1.2) was found on the contact between units UA3c/UA4 and in unit UB4c, and a smaller amount derives from unit UB5a (Giusti et al. 2018, 3).
Figure 2: a) the stratigraphy of the basin. The red triangle indicates the approximate location of 1 (Panagopoulou et al. 2015, 1); b) the stratigraphic columns of Area A and Area B of MAR-1, showing the sedimentary units, the relations between them and the absolute elevation. The red line indicates the contact between the find-bearing units (after Karkanas et al. 2018, 4).
1.1.2 Palaeontological and archaeological finds
Already since antiquity, the Megalopolis basin is known for its palaeontological finds, with the occurring giant bones being attributed to mythical creatures in ancient Greek writings (Konidaris et al. 2017, 1). The first systematic palaeontological excavations were carried out in 1902 by professor Skouphos and published in the 1960s by Melentis. However, this material lacks stratigraphic context and as a consequence cannot be used for biostratigraphic dating or other information on the Marathousa Member (Konidaris
- 11 -
et al. 2017, 1-3). Later investigations into the faunal remains from the Marathousa
Member were carried out by Sickenberg (1960s, published 1975), Benda et al. (published 1987), De Bruijn and Van der Meulen (1980s, unpublished) and Doukas, Theocharapoulos and De Bruijn (1995, published in Van Vugt et al. 2000) (Konidaris et al. 2017, 3; Van Vugt et al. 2000, 72-3). The studied assemblages contained fossils of large mammals like Mammuthus meridionalis, Hippopotamus antiquus, Stephanorhinus
etruscus and several cervids, smaller mammals such as Castor fiber, Sciurus cf. vulgaris
and mustelids, reptiles (a. o. Emys orbicularis and Natrix sp.) and microfauna including fish, birds, amphibians, three vole species (Pliomys aff. episcopalis, Mimomys
ostramosensis and Mimomys aff. savini) and the mouse Mus cf. spretus (Van Vugt et al.
2000, 72-8).
In the collection of fossils that Sickenberg studied, one specimen was not of faunal but of hominin origin: a tooth (Tourloukis 2010, 112). The tooth is a small upper third molar with an eroded crown. Its dimensions fall within the range of that of modern humans and Neanderthals, Heidelbergs and Homo erectus. The crown shape of the molar bears most similarity with earlier Homo taxa but cannot be assigned to a particular species with certainty (Harvati 2016 in Raksandic et al. 2018, 74). Although a species
identification is not possible, the molar does attest to hominin presence in the
Megalopolis basin during the time of deposition of the Marathousa member (Tourloukis 2010, 112).
From the site MAR-1 itself, a large number of large mammal bones and teeth were excavated in the field seasons of 2013-2016. The finds come from both Area A and Area B. In Area A, the skeleton of a Palaeoloxodon antiquus-elephant forms the main part of the faunal assemblage. The elephant bones were found disarticulated but in close proximity to each other (fig. 3a). This, the fact that there are no duplicate bones and that all bones are consistent in size and an ontogenetic advanced age, indicate that the bones indeed belong to a single individual animal (Konidaris et al. 2017, 4). Up to 2016, the skeletal elements that have been recovered are: the cranium with both upper third molars and the proximal part of the right tusk, 22 vertebrae including atlas, axis and sacrum, 26 ribs, the left humerus, the right ulna, the pelvis, the right femur, the right patella, the left tibia, 9 carpals and tarsals, 5 metapodials and 5 phalanges (Konidaris et
- 12 -
simplicidens, Hippopotamus antiquus and Cervus elaphus (Konidaris et al. 2017, 3). In
Area B, the large mammal assemblage is composed of isolated bones and bone fragments from Felis sp., Vulpes sp., Canis sp., Palaeoloxodon antiquus, Hippopotamus
antiquus, Bison sp., Dama sp. and Cervus elaphus (Konidaris et al. 2017, 3; Giusti et al.
2018, 2). Apart from large mammals, high numbers of fossils from micromammals, turtles, birds, fish, insects and terrestrial and aquatic molluscs have been found (Harvati 2018 pers. comm.; Panagopoulou et al. 2015, 5).
In close proximity and stratigraphic relation to the faunal fossils, large amounts of lithics were found in both areas (figs. 3a-d) (Panagopoulou et al. 2015, 6; Tourloukis and Harvati 2017, 2). The lithics are small-sized flakes and flake fragments, cores, some retouched tools and chunks or shattered pieces (fig. 4). Large Cutting Tools (LCTs) are absent. All technological traits of the lithic material point to a relatively uncomplicated operational sequence aimed at producing flake blanks (Panagopoulou et al. 2015, 5; Tourloukis and Harvati 2017, 2). Macroscopically identified use wear on retouched tools as well as on plain flakes suggests that no further modification of flakes was needed to be used for tasks like cutting (Tourloukis and Harvati 2017, 2). The small size of the lithics is probably a result of the properties of the used raw material. Most of the lithics are made from locally available radiolarite which occurs in small pebbles. The other materials that are used, flint and quartz, also occur in small pebbles. On top of that, the raw material is of mediocre quality, with many cleavage plains and impurities
(Tourloukis and Harvati 2017, 2; Thompson et al. 2018, 9-10). The character of the lithic industry from Marathousa-1 (flake-based small tools, no LCTs) bears a strong
resemblance to that of other Central European Lower Palaeolithic sites such as Schöningen, Bilzingleben and Neumark Nord 3 in Germany, Rusko in Poland,
Vértesszölös in Hungary, La Polledrara in Italy and Dealul Guran in Romania (Rocca 2016; Tourloukis and Harvati 2017, 2).
- 13 -
Figure 3: the distribution of the palaeontological and archaeological finds from MAR-1, in a) unit UA4, b) unit UA3c, c) unit UB4c and d) unit UB5a (after Giusti et al. 2018, 3).
- 14 -
Both the fossil bones and the lithics are extremely well preserved, with no evidence of rolling and no traces of rounding. This indicates a low-energy depositional environment and/or quick burying of the objects, which is consistent with the conclusion from the sedimentological investigation that MAR-1 is located at the shore of a palaeo-lake (Konidaris et al. 2017, 13; Panagopoulou et al. 2015, 6). The excellent preservation made the identification of taphonomic modifications on several bones possible. Modifications were found on bones originating from both excavation areas (Konidaris et al. 2017, 13). In Area A, cut marks were identified on an astragalus and a tibia of the elephant. The place and type of the cut marks suggest hominin exploitation of the elephant carcass for fat and meat (Konidaris et al. 2017, 13). In Area B, modifications were found on bones from different species. An elephant rib fragment shows cut marks possibly indicative of cleaning of the bone in the early stages of butchering. On bones from other species (fallow deer and a medium-large mammal) cut marks and percussion marks occur (Konidaris et al. 2017, 13-14). That the animal resources at the site were not only exploited for food, but also for raw material, is indicated by the presence of bone flakes and a bone shaft fragment with scars of flakes and impacts resulting from bone
knapping (Konidaris et al. 2017, 17).
The main finds layers in which the bones and lithics are found, are units UA3c and UA4, and UB4c and UB5a. The elephant skeleton lies on top of unit UA4 and is covered by unit UA3c. Most of the non-elephant bones from Area A are found inside sediments of unit UA3c (Giusti et al. 2018, 3). In Area B, the majority of the finds derives from unit UB4c while a small amount of bones and lithics was recovered from unit UB5a (Giusti et al. 2018, 3). The main find bearing units represent secondary depositional events (dilute mudflows and hyperconcentrated flows, see section 1.1.1). This raises the question whether the spatial association of the lithics and the faunal remains is reliable. Results of analyses of the site formation processes show that the elephant carcass, the other faunal remains and the lithics were originally located on or close to the surface of units UA4/UB5a. They were then eroded and redeposited or reworked by the
hyperconcentrated flow(s) of units UA3/UB4 (Giusti et al. 2018). Although not enough is known of the ‘occupational’ surface of units UA4/UB5a to use the spatial pattern of the finds for inferences about the exact use of space by hominin behaviour such as
- 15 -
remains is reliable and must have resulted from the exploitation of fauna at the lake shore (Giusti et al. 2018, 15).
Figure 4: a selection of the lithics found in the 2013 field season. The scale bar is 5cm (after Panagopoulou et al. 2015, 6).
- 16 - 1.1.3 Age of the site
The palaeontological finds discussed above point to a Middle Pleistocene date for the Marathousa member. The large mammal assemblage indicates a date of between 900 and 400 ka (thousand years) (Konidaris et al. 2017, 11).
Attempts have been made by Van Vugt et al. (2000) and Okuda et al. (2002) to correlate the different lignite seams and detrital beds with the eccentricity curve of solar
insolation. Both interpret couplets of one lignite seam and one detrital bed as a complete sedimentary cycle spanning one interglacial and one glacial. Van Vugt et al. (2000, 82) recorded the Brunhes/Matuyama magnetostratigraphic boundary in the upper part of lignite seam I or just above it in the lower part of the following detrital bed. According to their astronomical tuning, the detrital bed between lignites II and III in which MAR-1 is situated, would correlate to MIS 16 (Van Vugt et al. 2000, 86). Okuda et
al. (2002, 9) obtained an ESR date of 370 ± 110 ka from fossil bivalve shells collected just
above the base of lignite IV. Using this date, Van Vugt et al.’s (2000) recording of the Brunhes/Matuyama boundary and their own pollen analysis results, they attribute the detrital bed between lignite II and III to MIS 14 (Okuda et al. 2002, 11-12). Another possibility, which is in accordance with the ESR date but fits the pollen results less well, correlates the MAR-1 detrital bed with MIS 12. Van Vugt et al.’s (2000) correlation of the bed with MIS 16 does not fit the ESR date nor the pollen results (Okuda et al. 2002, 12-13).
Recent investigation of the magnetostratigraphy of the Marathousa mine by Tourloukis
et al. (2018, 8) identified the Brunhes/Matuyama boundary within the lowermost lignite
I. Again, applying a full glacial/interglacial cycle to a couplet of one lignite seam and a detrital layer, the layer in which MAR-1 is found is assigned to MIS 12 and dated between 420 ka and 480 ka (Tourloukis et al. 2018, 10). A second option attributing the detrital bed to MIS 14 is not preferred because it does not fit with new radiometric dates obtained from the site (Tourloukis et al. 2018, 12). The radiometric dates were obtained through post-IR IRSL and ESR dating. Post-IR IRSL was used to date feldspars from the sedimentary sequence between the top of lignite II and the base of lignite III, resulting in an age of between ~400 ka and ~500 ka for the sedimentary sequence (Jacobs et al. 2018 in Tourloukis et al. 2018, 10). The material for ESR dating was taken from molluscs deriving from unit UA2 and came back with an age of ~400 ka. Five
- 17 -
subsamples of a cervid tooth found in UB4c provided an age of approximately 500 ka (Blackwell et al. 2018 in Tourloukis et al. 2018, 10-12).
Taking all these dates into account, a broad age of between 500 ka and 400 ka for Marathousa-1 and a correlation to MIS 12 seems most accurate. This makes it one of the oldest open-air site in south-eastern Europe and the oldest in Greece (Tourloukis and Harvati 2017, 2; Tourloukis et al. 2018, 15).
1.2 Research questions and relevance
The archaeological finds (see section 1.1.2) show that hominins at the site MAR-1 exploited the resources offered by their natural surroundings. Locally available stone was used to make tools for tasks such as cutting. Cut marks and percussion damage on several large mammal bones indicate that the faunal resources were used for food and raw material. This raises the question which other natural resources offered by the natural environment could have been exploited by hominins. Therefore the main research question is:
What was the potential for plant exploitation by hominins at the Middle Pleistocene site Marathousa-1?
To answer this main question, several sub questions need to be answered first: - What was the local vegetation of Marathousa-1?
- How did the local vegetation of Marathousa-1 develop over time? - How were plants used by Palaeolithic hominins at other sites? - What known uses do plants present at Marathousa-1 have?
The answers to the research questions will first of all help in understanding the palaeoenvironment of Marathousa-1. Data on the local vegetation can be used in combination with data on for instance regional vegetation, geology, micro- and
macrofauna and hydrology to paint a picture of the palaeoecology at Marathousa-1, of which hominins evidently were a part as well. Understanding the palaeoecology of MAR-1 is important, because, since it is one of the oldest open-air sites in south-eastern Europe, this can shed light on the habitat type that hominins first settling the region preferred. Secondly, not much is known about the use of plants in the Palaeolithic (Bigga
et al. 2015; see also sections 4.1 and 4.2). This research will hopefully provide a starting
- 18 -
leading to more awareness of the role that plants could have played in the lives of Palaeolithic hominins.
In the next chapter, the material and methods used to answer the research questions will be outlined. In chapter 3, the reconstruction of the local vegetation at MAR-1 is presented and the limitations of this reconstruction are discussed. Also, the methods used for making the reconstruction are reviewed. In chapter 4, the possible plant use in the Palaeolithic is researched using several available sources. Combining the general uses for plants in the Palaeolithic and the known uses of plants found at MAR-1, the potential for plant exploitation at Marathousa-1 is assessed. In the concluding chapter, the materials and methods used in this research and the results they provided are critically reviewed, and the answers to the research questions are summarized. Finally, suggestions for further research are given.
- 19 -
2. Materials and methods
Analysis of macro-botanical remains provides detailed information on the local vegetation of a site (Lowe and Walker 1997, 185). This method is therefore used to answer the first to research questions. For the assessment of Palaeolithic plant use, a literature review has been deployed. In this chapter, the used materials and methods will be described in detail.
2.1 Sampling in the field
The material for the macro-botanical analysis was retrieved from a north-facing section in Area A. The section was located at the southern margin of the area, immediately next to some of the excavated Palaeoloxodon bones. Six bulk samples of about 2L of
sediment were taken. Extraction of the bulk samples was done bottom-up, to prevent contamination of the section face by sediment falling down from higher up. The bulk samples were each put in a plastic bag. The bags were double sealed and transported to Leiden, where they were stored in a fridge at just above 0˚C.
2.1.1 Description of the section and the samples
The sampled section face was approximately 65cm long. It consisted of two faces: the main face of the lower part, and a back face at the top (fig. 5). The samples are numbered 1-6 top down. The depths of the samples are given between 0-60cm, with 0cm being the top of the section.
Sample 1 (0-10cm) was partially taken from the main face and partially from the back face. The sediment consists of organic rich, dark brown silt and fine sand. It is not sure if this sample belongs to sedimentary unit UA3b or UA3c (Field 2017 pers. comm.). Sample 2 is divided from sample 1 by a contact or desiccation crack which contained a large piece of wood. Sample 2 (10-20cm) consists of dark brown sediment rich in organics and mollusc shell fragments. There are many deformed reddish brown patches present in the main matrix (Field 2017 pers. comm.). Sample numbers 3 (20-30cm) and 4 (30-40cm) contain the same sediment as number 2, but without the shell fragments. In addition to
- 20 -
the reddish brown patches, this sediment also contains deformed intraclasts of grey clay. Samples 2, 3 and 4 derive from unit UA3c (Field 2017, pers. comm.).
The contact between sample 4 and sample 5 (40-50cm) is clearly visible because of the presence of large amounts of compressed wood and smears of reddish brown sediment (Field 2017 pers. comm.). This is the contact between units UA3c and UA4 on which the elephant remains were found (see section 1.1.2). Samples 5 and 6 (50-60cm) are taken from fine grained dark brown, organic rich sediment. In the matrix some deformed grey clay intraclasts are present, but no shell fragments nor patches of reddish brown sediment (Field 2017 pers. comm.).
Figure 5: the sampled section. The numbers correspond with the numbers of the samples. The samples are divided by the lines. The red line indicates the contact between units UA3c and UA4, on which the elephant remains were found. The white line indicates a contact or desiccation crack that possibly separates units UA3c and UA3b. The scale bar is 50cm and the elevation of the black diamond 349.48masl (Field 2017 pers. comm.).
- 21 - 2.2 Data acquisition in the lab
The subsequent analysis of the material was performed in the Laboratory for Archaeobotanical Studies of the Faculty of Archaeology, Leiden University.
The bulk samples were subsampled to 200cc each using a water-filled glass cylinder of 500ml. First, the cylinder was filled with 200ml of water and then sediment was added till the water column in the cylinder had risen to 400ml. The volume of 200cc for every subsample (from now on simply called sample) was chosen on the assumptions that 1) only very little undiagnostic soft plant material would be preserved due to the old age of the sediments, and 2) also the amount of identifiable botanical remains would not be massive because of the age of the material. The remainders of the bulk samples were left in their plastic bags, sealed again and put back in the fridge. The samples were put in glass beakers with water and left at room temperature to soak, in order to break down the sediment. Washing up liquid was added to speed up the process1.
After soaking, the samples were sieved using a series of four consecutive sieves with mesh sizes of 1mm, 500μm, 250μm and 150μm respectively. The sieving was done with lukewarm tap water. All sieves were cleaned with tap water and a brush after use with each sample, to avoid cross-contamination of the samples. The four resulting size fractions per sample were put in a glass beaker each. Water was added to prevent the material from oxidising. The beakers were put in a bucket closed with lid and stored at room temperature for a week, after which a second round of sieving was performed. The largest fraction of all samples was left to soak even longer and subsequently sieved twice more, because big blocks of sediment were only slowly breaking down.
Of sample 1, the smallest fraction was re-sieved once because of the still high amount of sediment and the poor sorting of the material after the regular sieving. The mesh sizes of the sieves used for the re-sieving were 1mm, 425μm, 212μm and 150μm.
Recovery of identifiable macro-botanical remains such as seeds, nutlets, endocarps, sporangia and megaspores from the samples was done by putting a small amount of the sample in a petri dish with water and then visually scanning the dish using a
microscope. The used microscope is a binocular with a magnification of 6.3 to 40 times.
1Adding washing up liquid alters the isotopic make-up of the material. Nevertheless, because no
stable isotope analyses will be performed on the material and because the age of the material is far beyond the scope of C-14 dating, it could be added without negative consequences.
- 22 -
Macro-remains were picked with small tweezers or very fine brushes, and put on filter paper in another petridish. A preservative liquid consisting of one-third water, one-third alcohol and one-third glycerine was added to prevent drying out of the botanical
material and to prevent the growth of micro-organisms (rotting). This procedure was repeated until all the material from each sample was scanned. The unidentifiable parts of the samples, made up of sediment, general botanical rubbish and faunal remains, were collected in a plastic box with lid and kept for possible future checking. Due to a lack of time, it was later decided to completely discard the waste material. All petri dishes holding picked remains were stored at room temperature when checked regularly, or put in the fridge when not checked for a period of at least a month. The picked macro-remains were identified using the Digital Seed Atlas of the Netherlands (Cappers et al. 2006), the reference collection of the Laboratory for Archaeobotanical Studies and with the help and expertise of Dr. M. H. Field.
Identification of wood fragments was done by E. E. van Hees BA. Additional literature for specific taxa is mentioned in chapter 3. Identification to species level is preferred but not always possible, so in several cases identification up to genus or family level was the most precise identification possible. The number of remains per taxon was tallied. After identifying and counting, the remains of each taxa were put in small glass tubes filled with preservative. The tubes are stored at the Laboratory for Archaeobotanical Studies and are available for future consultation. Charcoal and unidentifiable wood fragments were not counted but noted as present or absent. Any indeterminate fossils will be identified later and are therefore not included in the total amount of fossils.
2.3 Data analysis: semi-quantitative or concentration
Due to time constraints, not all six samples have been picked completely. A choice has been made to only complete the richest sample, which is number 1. Of the remaining five, three other samples were chosen to provide an even spread along the sequence. These samples are numbers 3, 5 and 6. They are not completed yet, but between about 50% and 80% of the material per sample is analysed so far. Also from sample 1, data were collected broadly halfway the process of picking. This leads to a problem where no full concentration data of all samples are available. However, a semi-quantitative analysis of the data was possible. The tentative numbers of fossils per taxa were
- 23 -
grouped in four main categories: absent (0 fossils), rare (1 fossil), common (2-10 fossils) and abundant (>10 fossils). This is of course less detailed than absolute concentration data, but more informative than mere presence/absence data. Therefore, it enables the assessment of general trends in the development of the flora. Furthermore, the
incomplete semi-quantitative data of sample 1 can be compared with the complete concentration data, allowing for an interesting methodological side-question: what is the balance between time investment and information gain after completion of more than half of the material when analysing a rich macro-botanical sample?
To further analyse the macro-botanical data, all identified taxa were grouped into five ecological categories: aquatic, waterside, grassland, woodland and broad or
indeterminate habitat. The main ecological tolerances for each taxon were taken from the Heukels’ flora (Van der Meijden 2006), www.floravannederland.nl and
www.ecoflora.org.uk.
With the semi-quantitative data grouped into ecological categories, a diagram was made using the software program Tilia© version 2.0.41 (Grimm 2015).
2.4: Plant use: literature review
For the reconstruction of Palaeolithic plant use, a literature review was deployed. Starting point was the research of Gerlinde Bigga into the potential for plant use at Schöningen (see section 4.1.2) (Bigga 2018; Bigga et al. 2015 and references therein). All additional literature used to reconstruct Palaeolithic plant use is mentioned in the text of chapter 4.
The next step in the research was to assess the usability of the plant taxa that were present in the samples from MAR-1. For this end, three main sources were used: the website Plants for a Future (www.pfaf.org), Bigga’s findings (Bigga 2018, 78-91; Bigga et
al. 2015, 94-97) and Melamed et al.’s list of edible species from Gesher Benot Ya’aqov
(see section 4.1.2) (Melamed et al. 2016, supporting information). Since these three sources together provide a comprehensive overview of usable plant properties, no further sources were consulted.
- 24 -
3. Results of the macro-botanical analysis
In this chapter, the results of the macro-botanical analysis are presented and interpreted. English and Dutch names of the mentioned plant taxa can be found in appendix 1.
3.1 Semi-quantitative results: general characteristics
The semi-quantitative results of samples 1, 3, 5 and 6 make it possible to discern the general characteristics of the vegetation at Marathousa-1 (fig. 6; table 2). Most taxa (40 out of 61) grow in or near water, but grassland and woodland taxa are present too. This indicates that MAR-1 was located at the edge of a body of fresh water. The water must have been still or slightly flowing, because the identified aquatic taxa present cannot grow in fast-flowing water (www.ecoflora.org.uk). The water was probably not very deep, since Nuphar lutea, Nymphaea alba (fig. 7a) and Scirpus lacustris require water of less than 3m deep (www.ecoflora.org.uk). These results support the conclusion from the geological investigations that MAR-1 must have been located at the shore of a palaeo-lake (see section 1.1.1). The relatively big number of taxa with an unclassified habitat is likely due to the fact that none of these are identified to species level. The genera and families in the unclassified category contain species that grow in various habitats. However, in general it can be said that no species of the unclassified taxa occur in water. They can, though, grow in diverse habitats, ranging from waterside to woodland areas.
Table 2: the number of identified plant taxa per sample and per habitat and the total number of identified plant taxa per habitat for all samples combined, according to the semi-quantitative analysis.
Sample Aquatic Waterside Grassland Woodland Unclass. Total
1 18 17 3 2 9 49
3 15 8 0 1 9 33
5 14 13 1 2 12 42
6 8 7 0 0 6 21
- 25 -
Figure 6: a diagram with the semi-quantitative results from samples 1, 3, 5 and 6. Charcoal and wood fragments are given as present or absent. The red line in the lithology indicates the contact between UA3c and UA4, on which the elephant remains were found.
- 26 -
The results per sample differ. Sample 6 shows very little diversity in plant growth. It contains the lowest number of taxa of all the samples. The taxa that were identified are mainly aquatic (8 taxa) or waterside (7 taxa). Grassland and woodland species are absent. Also the total number of fossils is low, with only Ranunculus subg. batrachium,
Typha sp. and Dryopteris sp. (fig. 7b) being common in the assemblage. The high
numbers of fossils from these species are explained by the fact that these plants are mass-producers of the found fossils (respectively achenes, seeds and sporangia). It does not necessarily mean that these plant were also common in the vegetation at the time of deposition.
Sample 5 is much richer than sample 6. The number of identified taxa is twice as high and the total amount of remains is also much higher. However, the relative distribution of the taxa over the habitat categories is broadly the same. Close to two-third of the taxa has a wet habitat, and the three grassland or woodland taxa are only accounted for by one fossil each. The general wet character of the vegetation in samples 5 and 6 can be a result of the sedimentological unit from which they derive: both samples were taken from unit UA4. This does, though, not explain the difference in the amount of fossils and taxa. One explanation could be that sample 5 was almost completed, whereas sample 6 was not. The remainder of the sample could potentially provide additional taxa and higher numbers of already identified taxa.
Sample 3 again shows a wet environment, but it seems to have been even wetter than in samples 5 and 6. The amount of aquatic taxa (15) is twice as high as the waterside taxa (8). Grassland species are absent and the only identified woodland species is Rubus
fruticosus, in the form of one seed. The aquatic taxa could be relatively more numerous
as a result of the geology: UA4 (samples 5 and 6) was deposited in the marsh area surrounding the lake, whereas UA3c (sample 3) was deposited at the edge of the lake itself (see section 1.1.1). This may mean that the water level of the lake was higher at the time of deposition of UN3c than it was during the deposition of UA4.
Sample 1 is the richest of all the samples. It yields both the most identified taxa and the most fossil remains. The semi-quantitative data collected more or less halfway during the picking of this sample show that there was a wet environment, comparable to that of samples 5 and 6. Although the number of taxa is higher than in those samples, the relative distribution of the taxa across the habitat categories is similar. The only
- 27 -
difference is that sample 1 contains relatively more grassland and woodland species and less species with an unclassified habitat. The occurrence of more dry ground species may be a result of a lower water level in the lake, when the lake shore retracted and allowed dry ground species to grow closer to the site. However, the absolute number of fossils from grassland and woodland plants is not large, so it might also be a result of the high total number of fossils in this sample: the more remains, the bigger the chance of encountering some fossils of rare taxa.
Figure 7: examples of identified macro-botanical remains. a) a fragment of a Nymphaea alba seed; b) a sporangium from cf. Dryopteris (Field 2017 pers. comm.).
- 28 - 3.1.1 Some remarks on certain species
Some identified species are interesting and provide more information than just the environment in which they grow. Azolla filiculoides is nowadays only native to Australasia and the Americas (Field 1999, 86). It became extinct in Europe during the late Middle Pleistocene (Godwin 1975 in Field 1999, 92), probably at the end of MIS 9 (Erd 1970 in Field 1999, 92). Because of the timing of its extinction, A. filiculoides is an important biostratigraphical marker. The presence of A. filiculoides at MAR-1 is in agreement with the proposed age of the site (see section 1.1.3).
Potamogeton distinctus (here identified as Potamogeton cf. distinctus) that is native to
the warm and temperate regions of East and Southeast Asia (Zhuang 2011). Its
occurrence in Europe dates to the Middle Pleistocene (Field et al. 2000, 261). The exact timing of its appearance and extinction in Europe is not clear.
All other identified taxa are native to southern Europe.
3.2 Vegetation change?
The results of all four samples show a wet environment, indicating that the site was located close to the palaeo-lake at all times. It is very hard to observe any important changes in the vegetation over time with certainty. It is also unsure if the observed differences between the samples mentioned above represent markedly different types of vegetation. This is a result of various problems: the properties of the sediment from which the samples were taken, the semi-quantitative nature of the dataset, and the subjective analysis of the data.
3.2.1 Properties of the sediment
The sediments from which the samples were taken, are all reworked after deposition or are of a secondary depositional nature (see section 1.1.1). UA4 was formed by
hyperconcentrated flows and was affected by currents and waves after deposition in the marsh area around the lake shore. UA3c and UA3b are a secondary depositional units formed by hyperconcentrated flows that were emplaced under water after deposition. Thus, the origins of the sediment cannot be traced exactly. On top of that, because of the reworking after deposition, any fine depositional layers that might have been in place were disturbed. Sediments deposited at different moments got mixed up with
- 29 -
each other, for instance by waves and currents resulting from storms or by big animals (e. g. the elephant) entering the lake margin to drink, trampling the sediment. The estimated sedimentation rate across the basin is ~21cm per thousand years (Van Vugt et
al. 2000, 86) and precisely at the site ~10cm or ~26cm per thousand years, depending on
the age model for the site (Tourloukis et al. 2018, 13). From these sedimentation rates it follows that the sampled section of 60cm represents several thousand years. If for instance an elephant walked into the lake margin to drink, it may have disturbed thousand years of sediment deposition. A second problem, apart from the properties of the sediment itself, is when the macro-botanical remains were trapped in the sediment. It could have happened before the deposition of the sediment in the lake, between deposition and reworking or even after the reworking of the sediment. This means that both the origin of the macro-remains and the timing of deposition are not necessarily the same as that of the sediment in which they are trapped. But considering that the hyperconcentrated flows that deposited the sediment into the lake margins were probably of (close to) local origin, the macro-remains that could have been present in the sediment at the time of deposition of the sediment also must have been of local origin. If the fossils were trapped in the sediment after deposition in the lake, they are definitely of local origin. But still, the development of the vegetation is hard to decipher from reworked sediments, especially because it is not entirely sure how strongly the sediment was reworked.
Because each sample is taken from a, sometimes arbitrary, layer of 10cm thick, the resolution is not high enough to observe fine changes in the vegetation, if those were even still present in the sediment at the time of sampling. However, thinner slices of sediment for each sample were not possible, because samples for macro-botanical analysis need to be big enough to provide enough data.
3.2.2 Semi-quantitative nature of the dataset
The dataset is still incomplete. Two samples, numbers 2 and 4, are not included. The other samples are not complete yet. Therefore, only a partial diagram with semi-quantitative data of incomplete samples could be made. With these data, not much more can be said than presence or absence of certain taxa and the relative amount of fossils per sample. Any differences between the sample could have been caused by
- 30 -
actual differences in macrofossil assemblage. Or, they could be a result of the varying degrees of completion of the samples. For instance, the big difference in tentative total number of fossils between samples 5 and 6, that originated from the same
sedimentological unit, might be a reflection of the percentage of the sample that is already picked.
Only by picking the entire sample and making a complete concentration diagram with all six samples can the uncertainty about the amounts of taxa and fossils be removed.
3.2.3 Subjective analysis
Analysis of the semi-quantitative data was carried out “by the eye”. No statistical analysis was performed. As a result, it is impossible to state whether the observed differences per sample are statistically significant. Therefore, it is for instance not sure if sample 3 indeed represents a wetter environment than the other samples, and if samples 5 and 6 represent a similar environment although sample 5 is much more diverse than sample 6. The reason for not performing a statistical analysis is that the data are incomplete. Performing an objective statistical analysis of incomplete data would give false certainty, because it builds on unsolid foundations.
Secondly, the division of the found taxa into the broad habitat groups is arbitrary, especially when it comes to grassland, woodland and unclassified habitats. The chosen allocation of all tree taxa (except Alnus which is definitely waterside) to the woodland category is based on the mere fact that they are trees. Trees in the identified genera, like Sambucus sp. and Betula sp., can not only grow in forests but also in relatively open environments, ranging from damp to dry ground (www.soortenbank.nl). Rubus
fruticosus is shade tolerant but can also grow in almost full sunlight
(www.floravannederland.nl). If these taxa were placed in the unclassified category instead of in the woodland category, the diagram would look very different and the analysis would have a different outcome.
3.3 Sample 1: a comparison between semi-quantitative data and concentration data Sample 1 is the only sample that is completed. This makes it possible to compare the semi-quantitative data of this sample obtained broadly halfway the picking process with the concentration data from the completed sample (see fig. 6 and appendix 2).
- 31 -
Sample 1 held a total of more than 4024 identifiable macro-botanical fossils. These fossils include complete specimens and fragments of for instance seeds, achenes, nutlets, sporangia, oospores, endocarps and wood fragments. The total number of taxa is 45.
3.3.1 Observed differences and possible explanations for them
A number of taxa is not present in the semi-quantitative data but is identified in the complete concentration data. Some other taxa were present in the semi-quantitative data but have been identified differently after completion of the picking.
The remains of Hippuris vulgaris (fruit), Myriophyllum spicatum (fruit), Nuphar lutea (seed), Hypericum sp. seed, Monocote (stem fragment), Musci sp. (capsule operculum),
Rumex sp. (nutlet) and Solanum dulcamara (seed) were identified halfway, but are not
present in the concentration data. This is most probably because these remains were put aside with the other indeterminate remains to be looked at after this research. A re-evaluation of all fossils after the completion of all six samples will likely re-identify some of these specimens, whereas other main remain indeterminate.
The seeds tentatively identified as Rubus fruticosus have been included in the Rubus sp. after completion. None of the seeds was complete so it was not possible to firmly identify them to species level.
The fossils of Potamogeton acutifolius have been re-identified as P. trichoides. The endocarps of these two species are very similar so attention is needed when identifying them. For the final identification, more time was taken than during the tentative identification halfway the picking process.
The achenes from Ranunculus subg. Ranunculus have been identified to species level and assigned to Ranunculus sardous. The achenes from Ranunculus sceleratus have been assigned to R. sardous too. So effectively, all achenes from species in the subgenus
Ranunculus have been identified as R. sardous.
The nutlet of Eleocharis sp. has been identified to the species E. palustris.
Seeds of Juncus sp. and caryopses of Poaceae were not identified halfway, but are present in substantial numbers (44 and 134 respectively) in the completed data. This is because the picking of the 250μm fraction, in which most of these remains were found, only started after the halfway identification. The occurrence of sclerotia of the fungus
- 32 -
Cenococcum geophilum only in the complete data is also explained by the order of
picking.
The nutlets of Cyperus longus were first probably included in the group of Carex sp. triogenous nutlet, but were identified to species level by comparison to modern specimens from the reference collection.
Some endocarps of various Potamogeton species were not preserved well enough to be undoubtedly identified to species level. Therefore, it was decided to only identify them up to genus level resulting in the presence of Potamogeton sp..
During the second half of the picking, one seed of Verbena officinalis was encountered. This species had not been found earlier in the process.
The identification of some bigger wood fragments lead to the addition of cf. Salix sp. to the botanical assemblage.
The biggest difference between the semi-quantitative data and the concentration data of sample 1 is the total amount of fossils. Some examples: the number of megaspores from Azolla filiculoides increased from less than 10 to 33 and those from Salvinia natans went up from around 50 to 389. The number of Sagittaria sagittifolia embryos rose from 1 to 8 and the number of Cyperus fuscus seeds increased from 1 to 66. The amount of
Characeae sp. oospores and cf. Dryopteris sp. sporangia increased the most, each from
around 300 to more than 1000.
3.2.1 Information gain vs. time investment
There is a delicate balance between the information that can be gained by picking an entire sample, and the time that needs to be invested to complete a sample. The lab work on sample 1 (picking and identifying) took an estimated 250 hours. It took so long partially because the students working on it were not very experienced in working with macro-remains, but also because the sample turned out to be very rich.
The comparison between the semi-quantitative data and the concentration data of sample 1 has shown that by completing the sample information was gained in several ways: 1) “new” taxa were added, 2) some fossils were identified to a more precise taxonomic level, and 3) the total amount of remains increased markedly. Despite this, reconsidering the tentative identifications has resulted in some apparent “loss” of information, because several fossils were re-evaluated as being indeterminate or
- 33 -
identified to genus instead of species level. The indeterminate specimens will be studied later, after which they can be included again in the concentration data.
The information gain of completing the sample compared to the data from halfway is substantial. The “newly added” taxa and better identified taxa provide a more detailed picture of the local vegetation. The concentration data can be used to make a
concentration diagram of the entire sampled sequence once the other five samples are completed as well. It is no problem that some fossils were dubbed indeterminate and left to be looked at again at a later point. Careful identification of hard-to-identify specimens will only benefit the interpretation of the dataset. So completing the sample was worth the effort. If time is short, it is advised to either take a smaller sample (100cc) or to make sure that equal parts of each size fraction are picked.
However, it is an illusion that fully completing a set of samples gives an exact
representation of the vegetation present at the site. For instance, during the sorting of the sieving residues in the 2018 field campaign, several seeds of Celtis cf. australis were identified that originated from unit UA3c (Mike Field pers. comm.). The presence of
Celtis sp. At MAR-1 was already known from previous phytolith analysis, but the seeds
made an identification up to species level possible (Field 2018 pers. comm.).
3.4 Conclusions on the macro-botanical analysis
The macro-botanical analysis clearly shows that the environment of Marathousa-1 at the time of deposition was that of a lake shore. The majority of the identified plant taxa has an aquatic or waterside habitat and only small numbers of remains of few grassland and woodland taxa have been found. The used sampling, data acquisition and data analysis methods make it impossible to point out any important changes in the vegetation. All samples contained substantial numbers of macro-botanical fossils. Comparing the results of the semi-quantitative data and the concentration data of sample 1 made clear that although it takes a considerable amount of time, it is worth to complete samples in order to make a concentration diagram.
- 34 -
4. Discussion on Palaeolithic plant exploitation
Before it is possible to assess the potential for hominin plant exploitation at Marathousa-1, the use of plants in the Palaeolithic must be reconstructed. In this chapter, an overview of the available sources to reconstruct past plant use is presented. Drawing from these sources, the general plant use in the Palaeolithic is reconstructed. With this general view in mind, the flora present at Marathousa-1 as presented in chapter 3 is reviewed for its potential use to hominins.
4.1 Sources to reconstruct Palaeolithic plant use
The use of plants in the Palaeolithic is hard to prove due to the usually poor preservation of botanical remains at Middle and Late Pleistocene sites (Bigga et al. 2015, 93).
Therefore other sources are needed as well to reconstruct Palaeolithic plant use.
4.1.1 Information from non-botanical sources
Several non-botanical sources are useful in researching past plant use (e. g. Hardy 2010). The list of sources mentioned here is not exhaustive and includes only the most used sources: analogies with modern hunter-gatherer groups and modern non-human
primates, analysis of hominin bones and dental remains, and use-wear analysis on tools. Modern hunter-gatherer groups have been well studied and documented by
anthropologists, because their lifestyle is regarded to be an analogue for past human and hominin life ways before the adoption of agriculture ca. 10 ka ago (Marlowe 2005, 54). In the past decades, doubts about this assumption have arisen (Marlowe 2005, 54). But although modern Holocene hunter-gatherers have contact with agricultural
societies, probably inhabit different habitats than Pleistocene hominins (and humans) did and most importantly, have a far more complex technology than Pleistocene
non-sapiens hominins, they can be an important source of information to reconstruct the
evolution of human behaviour (Marlowe 2005, 65). In the lives of modern hunter-gatherers, plants play an important role. Take for instance the Ju/’hoansi (also known as the !Kung) from the African Kalahari desert: about 100 species of plants make up 85% of the groups diet, while only the remaining 15% derives from hunted game (Kelly 2014,
- 35 -
1112). Even groups that rely highly on hunting, such as the Aché in Paraguay, collect several different plants to eat, especially ‘on the go’ during hunting trips (Hawkes et al. 1982). Apart from diet, a substantial part of modern hunter-gatherer technology is also plant-based. The Ju’/hoansi use wooden digging and throwing sticks and bows, and arrows mainly made from cane. Metal adzes are fitted in wooden handles and arrows are carried in a quiver made from bark (Kelly 2014, 1112). The Nuvugmiut in northern Alaska too use (partially) wooden tools for purposes including hunting, cooking and storage (Kelly 2014, 1114-1115). Fire of course is made with plant resources too. The diet of great apes consists almost exclusively of plant parts like fruits, nuts and leaves, but is sometimes supplemented with food from faunal sources such as invertebrates and honey (Haslam 2013, 26). Chimpanzees are even known to target hard-to-obtain plant foods that need tools to access them. These plant parts include hard shelled nuts that need cracking, palm pith that gets pounded to produce edible pulp and underground storage organs (USOs) that are dug up (Haslam 2013, 26). The objects used to obtain these foodstuffs are made from stone or from plants. Different plants and parts of plants are used by chimpanzees as tools. Stalks, twigs and bark for instance are used in termite-fishing, leaves and fruits get used as water sponges and sticks are used for digging up USOs (Haslam 2013, 28). An important but often
overlooked aspect of plant exploitation by great apes is the construction of nests from branches and leaves, which can have important inferences for early hominin plant use but is archaeologically almost invisible (Haslam 2013, 30).
Apart from analogies, non-botanical archaeological finds can indirectly provide clues about the plant use of Pleistocene hominins. Nitrogen isotopes analysis of hominin bones can tell if the protein intake from an individual in the last decade of its life derived from animal or plant sources (Hardy 2010, 662). There are, though, some problems with this method, because the distribution of nitrogen isotopes at the base of food webs and the incorporation of dietary nitrogen into body tissues is not well understood yet (Makarewicz and Sealy 2015, 148). Additionally, plants contain very little protein and therefore already a small amount of meat in a diet dominates the nitrogen isotope ratio. It is therefore not possible to accurately determine the actual percentage of plant protein in a diet (Hardy 2010, 662-3). Microwear analysis of teeth provides information on the grittiness of the stuff which individuals chewed on. Plants are more abrasive than
- 36 -
meat and as a result leave more and different kind of striations on tooth enamel than meat does (Henry 2010, 20). Different studies into the dental microwear of
Neanderthals came back with varying results. Lalueza et al. (1996, 384) found that most Neanderthal specimens they studied must have had a predominantly carnivorous diet, with only some exceptions which had a mixed hunter-gatherer diet. On the other hand, Pérez-Pérez et al. (2003, 507-508) conclude that Neanderthals had a varied and highly adaptive diet consisting of both meat and plant foods.
Whereas stable isotope and dental microwear studies can only tell if and sometimes to
what degree plant foods were consumed, the analysis of phytoliths and starch grains
trapped in dental calculus can discern what taxa were being chewed on (Henry 2010, 31). Dental calculus analysis from Middle Palaeolithic Neanderthal remains has shown that throughout their habitat range they ate grass seeds, USOs and legumes of Triticeae, Fabaceae and Liliaceae (Power et al. 2018, 38-39). Commonly used grass seeds and USOs were supplemented with local plants such as date palms in the Near East and particular but unidentified uncommon USOs (Henry 2010, 191).
Phytoliths and starch grains are not only preserved in dental calculus, but also on the surfaces of stone tools (Henry 2010, 34). When found on tools, these microbotanical remains not necessarily provide evidence for the consumption of plant foods, but they do indicate the processing of plant parts (Henry 2010, 52-53). Since the goal of
processing plants is often to improve the edibility (Henry 2010, 52-53), phytolith and starch grain analysis on tools can provide information on plant foods in the absence of dental and skeletal material. The use-wear on stone tools too can give valuable insights into the processing of plants. For instance, the cracking of nuts leaves characteristic depressions on stones (Goren-Inbar et al. 2002, 2455). Nut cracking resulting in so-called pitted stones is observed in chimpanzee behaviour, in modern human hunter-gatherer societies and in the archaeological record (Goren-Inbar et al. 2002, 2455-2456).
4.1.2 Information from preserved botanical remains
In the rare cases where botanical (macro)remains are preserved at Palaeolithic sites, direct inferences about hominin plant exploitation can be made. Here, three different Palaeolithic sites at which plant remains were found are presented: the Acheulian Early-Middle Pleistocene lake shore site Gesher Benot Ya’aqov in Israel (e. g. Melamed et al.
- 37 -
2016), the Middle Pleistocene lake shore site Schöningen in Germany (e. g. Bigga et al. 2015) and the Middle Palaeolithic to Neolithic cave Theopetra in Greece (e. g. Tsartsidou
et al. 2013). All three sites hold diverse evidence for plant exploitation by hominins.
Gesher Benot Ya’aqov (GBY) has an age of ca. 780 ka and is correlated to MIS 19 (Goren-Inbar et al. 2000, 944-945). The site is known because of its location in the Levantine Corridor en route to Eurasia from Africa (Goren-Inbar et al. 2000, 944) and because of the evidence at the site for the earliest hominin controlled use of fire outside of Africa (Goren-Inbar et al. 2004). Waterlogged conditions resulting from the site’s location at the shore of palaeo-lake Hula resulted in the excellent preservation of macrobotanical remains (Melamed et al. 2016, 14674). A total of 117 plant taxa were identified, of which at least 55 edible plant species (Melamed et al. 2016, 14675). The edible parts from the identified species include nuts, USOs, fruits, seeds, leaves and stems (Melamed
et al. 2016, supporting information). Macrobotanical remains were recovered from both
geological and archaeological layers in equal numbers throughout the site’s sequence, but the found taxa differed between the geological and archaeological layers: in the archaeological layers, “the probability of finding items relating to key food plants (staples) was significantly greater” (Melamed et al. 2016, 14676). This suggests that hominins brought edible plants to the site on purpose (Melamed et al. 2016, 14677). Concluding, evidence from GBY indicates that Early-Middle Pleistocene hominins used a wide array of plant species for food and fire.
The site of Schöningen is found in a recently abandoned open cast lignite mine and is around 300 ka old. Its location at the shore of a palaeolake allowed for rapid
sedimentation and waterlogged conditions, leading to the preservation of diverse organic material (Bigga et al. 2015, 92). The site is known for the discovery of wooden spears in 1995 (Thieme 1997, 807). The spears were fashioned from fir, Abies alba, and spruce, Picea sp. (Thieme 1997, 808-809). Numerous other artefacts made from spruce and fir such as a digging stick and objects dubbed a throwing stick (“Wurfstock” or “Wurfholz”), a roasting spit (“Bratspieß”) and clamp shafts (“Klemmschäfte”) have been found too (Bigga 2018, 181-183). These clear artefacts show the use of plant resources as raw material for tools, but no indication of hominin use of plant material for food or fire is discernible at the site (Bigga 2018, 197; 217). However, the potential for plant exploitation at Schöningen was high, with edible parts of plants available year round
- 38 -
(Bigga 2018, 79-82). In short, evidence from Schöningen shows that Middle Pleistocene hominins used plants as raw material for tools and had the opportunity to eat plants in every season.
Theopetra is a cave site located in the mainland of Greece, 100m above the Thessaly plain and overlooking the Peneios River (Valladas et al. 2007, 303). In the cave, a
continuous sequence of sediments with a total thickness of 6.4m is found (Valladas et al. 2007, 303). The earliest occupation of the cave took place around the transition from MIS 6 to MIS 5, indicated by a radiometric TL date of 150 ka from the oldest
archaeological layer in the cave (Valladas et al. 2007, 306). Phytolith analysis of
anthropogenic combustion features (hearths) shows that hominins brought a variety of plants into the cave (Tsartsidou et al. 2013, 176). Plants exploited by hominins at Theopetra include grasses, Cyperaceae and the dicotyledonous fruits of Celtis sp. and
Lithospermum arvense. Use of these plants for food is most likely, although the high
amounts of the seemingly inedible L. arvense are puzzling (Tsartsidou et al. 2013, 180-181). Macrobotanical remains also suggest that several different seeds, legumes and fruit producing species were used or eaten in the cave (Kotzamani 2009 in Tsartsidou et
al. 2013, 181). Charcoal analysis shows that a various array of tree species was used for
fire (Ntinou and Kyparissi-Apostolika 2016). All the results together indicate that a wide array of plant species was exploited at Theopetra cave by hominins in the Middle and Late Palaeolithic, both for food and fire.
4.2 Hominin plant use in the Palaeolithic
Combining all the sources mentioned in section 4.1, three broad categories for plant use become apparent: food, raw material and fire. These categories are further explained in the following sections.
4.2.1 Food
All the sources presented in section 4.1 indicate that Palaeolithic hominins must have eaten plants: modern hunter-gatherers and primates (primarily) eat plants, evidence for consumption and processing of plants is found on dental remains and stone tools, and remains of edible plants have been found at archaeological sites.
