The Lower Palaeolithic site of Marathousa 1, Megalopolis, Greece: Overview of the evidence

Abstract Marathousa 1 is a Lower Palaeolithic open-air site located in the Megalopolis basin, an area in Southern Greece known for its fossiliferous sediments. Mining activities in the basin uncovered a thick sequence of Middle Pleistocene lacustrine deposits representing the environment of a palaeolake. Marathousa 1 was discovered in 2013 during a targeted palaeoanthropological survey and excavated subsequently by an interdisciplinary team from the Ephoreia of Palaeoanthropology–Speleology of Greece and the University of Tubingen, Germany. This article presents results from the ongoing investigation and reviews the state of knowledge about the site. Systematic excavations during five field seasons have exposed a total of 72 m2 and revealed a partial skeleton of the elephant Palaeoloxodon antiquus and remains of other large mammals in spatial and stratigraphic association with a “small tool” lithic assemblage. Faunal and taphonomic studies indicate the presence of cut-marks and percussion damage on elephant and other large mammal bones. The study of site formation processes, together with taphonomic and geostatistic spatial analyses confirm the association of fossil and hominin activity remains and the stratigraphic integrity of the site. Radiometric dating, geological and biostratigraphical evidence suggest that hominin activity at the site occurred between 0.5 and 0.4 Ma. Marathousa 1 is the oldest currently known archaeological site in Greece and the only Lower Palaeolithic butchering site in the Southern Balkans. It is also a key site for documenting high resolution palaeoclimatic, palaeoenvironmental and cultural records of a geographical area that potentially acted as a refugium during the successive waves of hominin colonization of Europe.


Introduction
In the last decades, evidence of the earliest occupation history of Eurasia has changed significantly (Bar- Yosef and Belfer-Cohen, 2001;Dennell, 2003;Dennell and Roebroeks, 2005;Moncel, 2010;Moncel et al., 2018) with the presence of hominins attested at 1.8 Ma at Dmanisi in Georgia (Agustí and Lordkipanidze, 2011 and references therein) and 1.4 Ma at 'Ubeidiya (Bar-Yosef, 1989;Bar-Yosef and Goren-Inbar, 1993), 0.78 Ma at Gesher Benot Ya'aqov in Israel (Goren-Inbar et al., 2000) and at around 1 Ma in Turkey (see Dinҫer, 2016 for a summary of the recent evidence). In Europe, the earliest traces of hominins come from the Mediterranean area which probably hosted a first dispersal event and served as a core area for a range of expansions of hominin populations northwards (Dennell et al., 2011) and are dated to over a million years ago (Carbonell et al., 2008;Arzarello and Peretto, 2010, but see Roebroeks, 2001;Muttoni et al., 2010;.Early dates around 1.6-1 Ma that need to be verified have been recently suggested for Eastern Europe (Ivanova, 2016), although secure and dated Lower Palaeolithic contexts are rare in this area (cf. Doboş. and Iovita, 2016). Despite the fact that Greece lies at the crossroads of Europe, Africa and Asia and, therefore, possibly at the most direct pathway in the movements of hominins between Africa and Eurasia, there is a scarcity of evidence for hominin presence in the Early and Middle Pleistocene, due to preservation biases and the dearth of systematic archaeological research (Harvati et al., 2009;Tourloukis, 2010;Harvati, 2016;Tourloukis and Harvati, 2018). This paper presents recently obtained Lower Palaeolithic evidence from the excavation of

Geographic and geological background
Marathousa 1 is situated in the basin of Megalopolis (Fig. 1), an area in southern Greece long-known for its fossiliferous sediments (e.g. Melentis, 1961). The Megalopolis basin is an inter-mountainous depression that was formed during the Late Miocene -Pliocene as a result of extensional tectonic movements. During parts of the Pleistocene the basin hosted a large shallow lake. The sedimentary sequence of the basin includes lacustrine and fluvial deposits that reach a thickness of more than 250 m and are divided stratigraphically into six formations (Fm). The formation of interest is the "Choremi Fm" (Fig.1), which dates to the Early and Middle Pleistocene (Vinken, 1965;van Vugt et al., 2000) and is subdivided into the Marathousa and Megalopolis Members (Mb). These members include lacustrine and fluvial sediments, respectively, and represent glacial-interglacial alternations (Nickel et al. 1996;van Vugt et al., 2000;Okuda et al., 2002).
Since 1969 mining operations at the open-cast lignite mine of Megalopolis have exposed thick sequences of deposits, enhancing the visibility of fossiliferous horizons. Previous work indicated that the area is potentially important for palaeoanthropological/palaeolithic research (Lénormant, 1867;Sickenberg, 1976;Darlas, 2003;Harvati et al., 2009). However, systematic research targeting Pleistocene sites had never been conducted. In 2012 a systematic interdisciplinary surface survey was initiated by a joint team of the Ephoreia of Palaeoanthropology-Speleology of Greece and the University of Tϋbingen (Germany) in the framework of the ERC project "Palaeoanthropology at the Gates of Europe" (PaGE), with the aim to discover primary context Pleistocene archaeological sites (Harvati and Tourloukis, 2013;Panagopoulou et al., 2015;Harvati, 2016). The survey discovered numerous palaeontological (Athanassiou et al., this issue) and palaeolithic findspots, the latter attributed mainly to the Middle Palaeolithic (Thompson et al., this issue). Since the focus of research has been the in situ recovery of secure, stratified and datable contexts, the survey concentrated on the examination of the sections exposed by mining activities. Marathousa 1 was discovered in 2013 when members of the research team identified the first Palaeolithic artefacts in association with large mammal remains in an exposed profile (Panagopoulou et al., 2015). Marathousa 1 is located in sediments of the Marathousa Mb, which consists of lacustrine clay, silt and sand beds with freshwater bivalves and ostracods, and thick lignite seams. The site is located at 350 m above mean sea level (masl), inside the Marathousa lignite mine at 5 the center-west part of the basin (Panagopoulou et al., 2015). The cultural and fossil material was found in the detrital unit, which is sandwiched between two lignite seams ( Fig.   2, and see below, Stratigraphy section).

Excavation: goals, methodology and results
During the Marathousa mining activities, the area was formed into artificial terraces. When the site was located, the find-bearing horizon was covered by ca. 4 m. of sediments, which were removed mechanically before the start of the archaeological investigation. In the same year a salvage excavation was initiated which developed into a long term systematic research project (Panagopoulou et al., 2015). Here, we report the first results of the ongoing excavation.
Fossiliferous and cultural deposits were observed throughout ca. 100 m along the exposed section of the lignite mine (Thompson et al., this issue). Results of the rescue excavation suggested that an in situ Middle Pleistocene horizon was preserved along the western bank of the palaeolake. A grid system of 1×1 m units was set up and trenches were opened in two Excavation Areas, A and B (Fig. 2) to investigate the spatial distribution and main concentrations of finds and to locate the extent of the site. Due to the nature of the finds, excavation was conducted in 1×1 m squares and 10 cm spits in Area A, where excavation focused on recovery of the elephant remains; and in sub-square quadrants of 50×50 cm and 5 cm spits in Area B, where a denser lithic assemblage associated with faunal remains was found. Due to the compaction of the sediment and the high organic content, sediments were excavated with leaf trowels and small knifes, by stratigraphic unit, following the geological layers and/or the presence of archaeological features. In close proximity to the fossil finds wooden tools were used, in order to avoid damaging the surfaces of the bones or imitating the effects of cut-mark on the fossils. A total station was employed to document three-dimensional coordinates of finds including all lithic artefacts, teeth, bone and organic material ≥ 2 cm), as well as micromammal and other micro-faunal and palaeobotanical remains of special interest, such as eggshells or seed/fruits, sediment samples and sedimentary features. Dense clouds of surface points of the elephant remains were recorded using both a total station and a close range photogrammetric technique (for a detailed account of the recording system, see Giusti et al., this issue). The dimensions, dip/orientation of finds were also recorded for fabric analysis (Giusti et al., this issue). The Total Station data were transferred to a GIS data base to three-dimensionally reconstruct 6 find distributions and occupational surfaces. Systematic water-screening of all excavated sediments through a 1 mm mesh was initiated from the beginning of the project to ensure the recovery of numerous small finds, including lithics, faunal (mammals, reptiles, fish, mollusks, ostracods, insects etc.) and botanical (fruits, seeds, wood, charcoal etc.) remains.
Flotation of selected samples for the recovery of macrobotanical remains was also implemented. Furthermore, samples were collected for micromorphological, microfaunal, microbotanical (including phytoliths) remains and radiometric dating analyses. Sampling procedures involved both a site-specific and an off-site approach (selected profiles of the open-cast mine were also sampled). The find-bearing stratigraphic units (see stratigraphy section below and Karkanas et al., this issue) were recovered in both excavation Areas, and were investigated over a total of 72 m 2 , accounting for 55 m 3 of excavated sediment.
In Area A, excavations were initially devoted to the collection and protection with plasterjackets of the fossil material that was in danger of eroding out of the artificial section profile.
The work focused on the cranium of an elephantid, as well as on a number of the axial skeleton elements, presumably of the same individual, that were exposed in the section.
Taxonomic identification based on the craniodental morphology attributed the cranium to the species Palaeoloxodon antiquus (Konidaris et al., this issue). The expansion of excavation units in subsequent seasons revealed numerous well-preserved elephant skeletal elements in close association in the same layer, including a humerus, an ulna, a femur, a tibia, ribs, vertebrae, the pelvis, autopodial elements and tusk fragments (Fig. 3). Their stratigraphic and close spatial association with the cranium and the subsequent study (Konidaris et al., this issue) indicated that they indeed belong to the same individual. The elephant remains were found in direct spatial and stratigraphic association with lithic tools and rich macroand micro-faunal and floral records, including seed micro-remains, wood and charcoal fragments, birds, reptiles, ostracods and coleopters. As suggested by the distribution patterns (Giusti et al., this issue), the lithics found in Area A are mostly spread around the elephant skeletal remains (Fig. 4).
Excavation Area B, located 60 m to the south of the elephant fossil accumulation along the exposed west section of the lignite quarry (Fig. 2), revealed an assortment of lithic and faunal materials distributed across well-defined archaeological layers. Additional elephant elements, including the proximal end of a tibia, vertebra, rib and tusk fragments, were found. Subsequent study of the material indicated that at least the tibia represents a different individual from the one found in Area A. A higher density of lithics and the remains of other large mammals (e.g. cervids, bovids, hippos, carnivores) were also recovered from 7 Area B, along with fragments of rotted organic material and a range of organic remains, including mollusks, ostracods and microvertebrates. As in Area A, the distribution of the faunal remains correlates closely with the distribution of stone tools (Fig. 4, Giusti et al., this issue). interglacial periods, while detrital intervals likely correspond to glacial periods (Nickel et al. 1996;van Vugt et al., 2000;Okuda et al., 2002;Siavalas et al., 2009).

Stratigraphy
The clastic sedimentary sequence is thicker in Area B and more compressed in Area A (Fig.   5), indicating that the latter was probably located closer to the lake shore. Correlation between the two excavation areas was accomplished on the basis of stratigraphic, sedimentological and geochemical evidence (Karkanas et al., this issue). Both excavated areas A and B are characterized by a lower clastic sedimentary sequence, devoid of cultural remains (UA4 to UA7 and UB6 to UB10 respectively). These in turn overlie lignite unit II (UA7 and UB10 respectively). Sedimentation in this part of the sequence consists of bluish organic-poor muds and sands and appears to be relatively continuous and deposited mostly under water. According to the sedimentological and micromorphological analysis the sediments were deposited in a shallow to very shallow water lake environment. Eroded 8 remnants of a bluish mud layer are capping the sequence in both areas. This mud layer is thicker in Area B (UB6) and is found as thin remnants in Area A (UA4).
The overlying sedimentary sequence (UA3-2 and UB5-2) follows a major depositional hiatus (UA3/4, UB5/6), which is attributed to exposure and erosional processes. The erosional contacts UA3c/4 and UB4c/5a separate the cultural units in both areas and are overlain by a series of mudflows and hyperconcentrated flows deposited at the fringes of the palaeolake. In summary, the context of Marathousa 1 represents a depositional environment close to the shore of the lake, punctuated by higher energy events, such as mudflows and periodic subaerial exposure with dry, terrestrial depositional conditions (Karkanas et al., this issue).
The cultural deposits are embedded in such a mudflow formed close to the shores of the palaeolake. Several uncomformities in the sequence indicate temporal gaps of unknown duration in the sedimentation and are related to fluctuating lake levels caused by short-term climatic variability.

2.3.2Chronology
As already discussed, the find-bearing layers occur between Lignite Units II and III; therefore, according to the chronostratigraphic model of van Vugt et al., (2000), Marathousa 1 dates to Marine Isotope Stage (MIS) 16, or to MIS 14, according to Okuda et al. (2002). While relative age information is sufficient as a first approximation, radiometric, fine resolution chronology is necessary for placing the cultural evidence from the site in the wider discussion of behavioural evolution processes in this time range. Thus, in order to refine the chronology we initiated dating of the deposits by Electron Spin Resonance (ESR) and post-infrared Infrared Stimulated Luminescence (post-IR IRSL) dating. Samples for cosmogenic nuclides analysis were collected and results are pending.
Preliminary Electron Spin Resonance (ESR) age determinations indicate that the age of Marathousa 1 should fall around 450-500 ka BP (Blackwell et al., 2016), although these results should be confirmed by the dating of more samples, currently underway. On the other hand, sediment samples from the two excavation areas were dated by post-infrared Infrared Stimulated Luminescence (post-IR IRSL) and provided age determinations between 400 and 450 ka (Fig. 5), consistent with deposition of sediments during MIS 12. (Jacobs et al., this issue). In order to place constraints on the age of the site, the magnetostratigraphy of the Megalopolis basin was investigated and correlated to a standard isotope record of Pleistocene climatic variability. According to the preferred interpretation put forth by

The faunal and cultural assemblages
Large mammals include a castorid, two mustelids, a felid, two canids, an elephantid, a hippopotamid, a large bovid and two cervids (Konidaris et al., this issue), while small mammals (e.g. the water vole Arvicola, Doukas et al., this issue), birds (Michailidis et al., this issue), reptiles, amphibians and fishes are well represented in the assemblage. As already discussed, of particular interest is the partial skeleton of the straight-tusked elephant Palaeoloxodon antiquus found in close spatial association in the same layer of Area A (Fig 3), which also contained several other faunal remains referred to beaver, otter, hippopotamus the presence of a "small tools" (sensu Burdukievicz and Ronen, 2003) lithic industry (Fig. 8).
The employed technological approach aimed at the production of blanks from minimally prepared cores (evidence for core preparation is lacking) and then the selection of flakes for use, either without secondary modification or with retouch to produce a notched or pointed edge. Secondary modification is also systematically used for the manufacture of backed edges, probably to facilitate manual prehension or hafting. The assemblage composition (rarity of cores and cortical blanks and dominance of small debitage and blank shaping byproducts, such as retouch/resharpening flakes, chips, and spalls) indicates that the later phases of the reduction sequence are dominant and that lithic activity at the site was mainly geared towards tool manufacture, use, maintenance and discard. Bifacial technology is not recorded so far and the assemblage is clearly not related to the Acheulean. Nevertheless, the absence of bifacial technology from the Marathousa 1 lithic assemblage is not conclusive evidence of its absence in the region (cf. Lénormant 1867, for a handaxe collected in the Megalopolis area in association with elephant remains), especially in the light of recent discoveries of Acheulean related assemblages on Aegean islands (see below, Discussion section).
Except for the lithic assemblage, Marathousa 1 has produced evidence of organic technology, rare so far in Greece, in the form of occasional bone flakes and tools and a bone percussor (Fig. 9)

Discussion
The preceding sections have presented evidence on the different aspects of the site: excavation, stratigraphy, fauna, palaeoenvironment and archaeology. In this section, the data generated by the preliminary results of the various specialist analyses are integrated to

The archaeological context of Marathousa 1
13 After five years of systematic excavation, geoarchaeological assessment of the site's formation processes and analyses of the faunal, palaeoenvironmental and cultural material, we have a working hypothesis of the site's character and the hominin activities reflected in the assemblages. In the archaeological horizon uncovered in Area A, a partial skeleton of a Palaeoloxodon antiquus spatially and stratigraphically associated with remains of hominin activity has been located, representing manufacture, use and maintenance of lithic tools.
Hominin presence is also testified by bones with cut-marks. Well-preserved botanical remains (including plant macro-fossils, wood, charcoal and phytoliths) and a wide range of macro-and micro vertebrates (bovids, cervids, carnivores, micromammals, turtles and birds) from the same stratigraphic context provide excellent proxies for the reconstruction of the  (Konidaris et al., this issue), the sedimentology of the cultural layers, as well as those immediately overlying and underlying them, indicate relatively low-energy site formation processes. This assumption is also supported by the mint condition of the stonetools (Tourloukis et al., this issue b), the bone surface preservation, which indicates rapid burial (Konidaris et al., this issue), the general scarcity and small size of any present gravels, the excellent preservation and rich variety of organic remains (spores, wood fragments, etc.) and fauna (including fish, reptiles, amphibians, micro-mammals, micro-fossils, coleopters) and the occasional bone refits (Konidaris et al., this issue). All the above data clearly indicate a minimally disturbed context.
To assess the stratigraphic context of the cultural remains and the taphonomic processes that affected their distribution within the excavated sediments and, hence, the integrity of the site and the validity of our behavioural interpretations, a set of spatial statistics, 15 including fabric, vertical distribution and point pattern analyses, was implemented in a geoarchaeological and taphonomic framework to the recorded distribution of remains in both excavated Areas. Results demonstrate an autochthonous deposition subject to localized minor reworking and a short transport distance in both areas (Giusti et al., this issue). It is suggested that in Area A the elephant skeletal remains laid on an ephemerally exposed surface (at the contact of UA3c/4, Fig. 5) and were relatively rapidly covered and only slightly moved by the mudflow deposit of UA3c (Karkanas et al., this issue). In the same vein, in Area B the fossil and cultural materials were most likely derived from the erosion of exposed mudflat areas and were re-distributed locally and buried by the same depositional event represented by unit UB4c (Karkanas et al., this issue). In this sense, hominin activity at Marathousa took place in the immediate vicinity of the banks of the palaeolake.
As is the case in open-air sites (cf. Hovers et al., 2014) and indeed in every archaeological site, the combined results of the analyses showed that both hominin activity and natural processes contributed to the formation of the archaeological record at Marathousa 1.
However, despite the slight local reorientation and displacement of remains, no major depositional processes have affected the cultural record, thus enabling behavioural interpretations.
Of interest is also the question of whether the site preserves a single episode of use for the butchering of animals, a time averaged palimpsest of different activities and durations and/or a palimpsest of natural and anthropogenic episodes, i.e. independent carnivore and hominin-created bone accumulations. In other words, does Marathousa 1 represent a temporally restricted, specific use of a place by hominins? Or a palimpsest of different natural and anthropogenic episodes in the area of the palaeolake at different times? Different lines of evidence, namely the much higher density lithic cluster, the number of species present and the higher bone fragmentation, as well as the spatial association of the faunal assemblage in Area B, suggest variations in the use of the two excavation areas of the site, that could be due to repeated visits in an area farther from the shore of the palaeolake, more conducive to several activities. In this sense, the assemblages in Area B could have been created over a short but currently unknown time-span (over which animal carcasses were still edible), in fact corresponding to an accumulation of multiple events and indicating partially overlapping but individual activity episodes.
On the other hand, with the information currently available, distinguishing anthropogenic from non-anthropogenic signatures and fine-tuning the nature and sequence of depositional 16 events is not possible. More data from the continuation of the excavation and the use of different approaches for deciphering between single and multiple accumulation events at different temporal/spatial scales (e.g., detailed taphonomic analysis, lithic and bone refitting studies, tooth isotopic variations), are needed to understand the site biography and the natural and human impacts on bone accumulations.

Marathousa 1 in context
The data regarding the Lower Palaeolithic in Greece is intermittent and fragmented both chronologically and spatially and, in most cases, the available information is rudimentary at different levels. With the earliest of fossils consisting of one cranium from the cave of Petralona and an isolated tooth from Megalopolis (see Harvati, 2016 for a review of the palaeoanthropological data), most of the evidence consists of lithics from surface scatters or shallow sub-surface concentrations and is, consequently, marred with unresolved problems of taphonomy and chronology (for a summary and critical review of the evidence see Tourloukis, 2010;Elefanti and Marshall 2015;Papoulia, 2017;Tourloukis and Harvati, 2018 and references therein). Nevertheless, knowledge pertaining to the Greek Lower Palaeolithic has considerably improved in recent years through the implementation of targeted surveys (Strasser et al., 2010;2011;Runnels et al., 2014), excavations (Panagopoulou et al., 2015;Galanidou et al., 2016) and re-evaluations  of open-air sites that provided important new evidence.
Lower Palaeolithic sites are scattered on the mainland and on Aegean islands (Fig.10). With the exception of the cave of Petralona, they are all open-air sites and findspots, a pattern well attested in the Eurasian Lower Palaeolithic, where the use of caves is only sporadically recorded (cf. Chazan et al., 2012). In terms of chronology, the Early Pleistocene is absent from the record and, except for Marathousa 1 which dates at 400-500 ka, in the few cases where radiometric dates are available (Strasser et al., 2011;Tourloukis et al., 2015;Galanidou et al., 2016), they indicate a late Middle Pleistocene age. The sites are associated with various depositional settings (marine terraces, tectonic depressions, fluvial and lacustrine deposits) and, as a result, reflect several taphonomic histories and potential for behavioural inferences. Faunal materials are either almost completely absent or not associated with material culture remains. As regards the lithic records, the assemblages contain both Large Cutting Tools (LCTs, sensu McNabb et al., 2004) with Acheulean affinities (Runnels et al 1993;Strasser et al., 2010;Tourloukis et al., 2015;Galanidou et al., 2016) and "small tool" assemblages (Panagopoulou et al., 2015), suggesting an important and, as yet, poorly understood variability in the cultural record that could contribute new and important data in the wider discussion of the spatio-temporal distribution of Acheulean versus non-Acheulean assemblages and the Out of Africa 2 hypothesis (cf. Gallotti, 2016;Rocca et al., 2016;Sharon and Barsky, 2016). Except for sporadic remains on the mainland, LCTs sites/findspots have recently been located on the near-shore island of Lesvos (Galanidou et al., 2016), and, most intriguingly, in view of the proposed southern pathway of early hominin migration into Europe (Muttoni et al., 2010)  Europe and Western Asia during this period (Roberts and Parfitt 1999;Dobosi, 2003;Barsky and de Lumley 2010;Ollé et al., 2013;Wenban-Smith, 2013;Aureli et al., 2016;Moncel et al., 2018). It is clear that a set of parameters (e.g. raw material, site function, removal of curated items from the record, social practices in isolated groups) affects lithic assemblage composition and invoking biological and cognitive change or waves of Out of Africa expansion (Sharon and Barsky, 2016; but see Gallotti, 2016) might be misleading in several cases. Middle Pleistocene hominins employed multiple technological solutions in different contexts, and LCts and "small tool" assemblages probably reflect differential distribution of archaeological signatures within a varied technological repertoire.
We have only started to recognize the complexity of the Greek Lower Palaeolithic record and, given the complex topography and the varied environments of the country, the lack of evidence for site functions and the absence of an established chronostratigraphic framework, further evidence is needed from well-preserved and dated contexts, displaying different environmental adaptations and functional characteristics, to disentangle the distribution of small vs Large Cutting Tool assemblages and to see if it is related to diverse ecological settings, subsistence strategies, sub-regional traditions and chronologies. With a stratigraphically intact and radiometrically dated context and a gamut of in situ environmental proxies that would help establish correlations between local environment and hominin activities, Marathousa 1 represents such a case. The site will undoubtedly serve as a reference point for building a chronostratigraphic framework for the Lower Palaeolithic  -Labin, 2011;Abbate and Sagri, 2012) and was a period of great importance for both biological and cultural aspects of human evolution (e.g. Roebroeks, 2001;Stiner et al., 2009;Barham 2010;Harvati et al., 2010;Arsuaga et al., 2014;Gamble et al., 2014;Stringer 2016;Hublin et al., 2017;Posth et al., 2017). A focal research question is the type of subsistence strategies hominins followed during this time, as they dispersed from Africa across a highly variable Eurasian landscape. In this context, hominin exploitation of megafauna and especially hominin-elephant interaction is a topic that has attracted major attention in recent years (e.g. Ben-Dor et al., 2011). However, the co-occurrence of proboscidean remains and lithics in several sites does not necessarily indicate exploitation of megafauna whether hunted or scavenged (cf. Freeman, 1994;Villa et al., 2005;Domínguez-Rodrigo, 2008). In some of the sites of the period, direct evidence of anthropogenic processing of faunal remains is either non-existent, or is obscured due to the impact of taphonomic and post depositional processes on the distribution of the fossil and cultural remains, thus limiting behavioural inferences. The discussion below focuses on a small but increasing number of Middle Pleistocene sites from the Mediterranean area ( Fig. 11) with clear evidence of hominin interference on skeletal material in the form of butchering or other kinds of carcass processing, pointing to important changes in subsistence behaviour.
Although the above sites preserve a strong and unquestionable anthropogenic signal, it is still not clear if hominin exploitation is related to hunting and/or scavenging practices.
Hunting has been suggested at Gesher Benot Ya'aqov (Goren-Inbar et al., 2000) in the Mediterranean, at the late Lower Palaeolithic sites of Schöningen (Starkovich and Conard, 2015;van Kolfschoten et al., 2015) and Lehringen (Veil and Plisson, 1990) in Germany and hypothesized at the Ebbsfleet site (Wenban-Smith, 2013) in the U.K., while in most other cases it seems that the bone remains represent the result of a systematic or occasional exploitation of already dead animals (cf. Boschian and Saccà, 2015). Be that as it may, it is becoming increasingly obvious that proboscidean exploitation data indicate a more than a marginal strategy of protein and fat obtainment on the part of hominins (Yravedra et al., 2010;Ben-Dor et al., 2011).
Like Marathousa 1, most Middle Pleistocene proboscidean butchering sites from the Mediterranean are related to permanent or ephemeral water bodies, where elephants likely gathered during droughts. The repeated occurrence of Lower Paleolithic sites with proboscidean remains in lakeside environments in the Mediterranean and further afield suggests that the hominin groups were well aware of the movements of migrating herds, selected locations in the landscape which provided the best opportunities for accessing them (cf. Devѐs et al., 2014) and practiced regularly on-site butchery activities for meat procurement. In terms of cognitive and behavioural complexity, this pattern suggests the ability to share information on the environment and the resources available (cf. Coward, 2016) and organize a successful food acquisition strategy through social interactions and strategic planning (Roebroeks, 2001). (c) They integrate into the framework of similar contexts from the Middle Pleistocene archaeology of Eurasia, enabling the examination of the evidence in the wider discussion of behavioural evolution processes in this time.

Conclusions
Marathousa 1 is not a single site, but rather part of the archaeological landscape of the Megalopolis basin, now largely destroyed by time and mining activities, but still preserving thick Pleistocene deposits. From the results of our survey (Thompson et al., this issue), it is clear that the basin was a place that hominins used repeatedly, forming a focus of activity for at least 400-500 ka and making it possibly one of the richest Palaeolithic archaeological landscapes in the Southern Balkans. In addition to the continuation of the excavation, future work will concentrate on the reconstruction of the complex trajectory of landscape evolution and on locating more stratified Pleistocene localities. The aim is to examine how hominin populations and their behavioural repertoire adapted to a changing landscape within a single region through deep time.        Lithic artefacts from 6,7,9,[11][12][13]15,[17][18][19]21,24: flakes,5,8: chips,10,19,20: composite tools,4,14: retouched pieces, 16: backed piece, 18, 21-23: backed knives, 24: core. Numbers 4, 12, 17, 18 are fragments. The lithic material is stored at the Ephoreia of Palaeoanthropology-Speleology in Athens.

Figure 9
Bone artifacts from Marathousa 1: percussor on a diaphysis fragment of probably an elephantid limb bone (upper row); flake (lower left); flake with denticulated edge (lower right). The osseous material is stored at the Ephoreia of Palaeoanthropology-Speleology in Athens.