Accepted Manuscript Title: Site formation processes in caves: the Holocene sediments of the Haua Fteah, Cyrenaica, Libya Authors: Chris Hunt, John Davison, Robyn Inglis, Lucy Farr, Tim Reynolds, David Simpson, Graeme Barker, Hwedi el-Rishi PII: S0305-4403(10)00026-9 DOI: 10.1016/j.jas.2010.01.021 Reference: YJASC 2417 To appear in: Journal of Archaeological Science Received Date: 11 September 2009 Revised Date: 8 January 2010 Accepted Date: 14 January 2010 Please cite this article as: Hunt, C., Davison, J., Inglis, R., Farr, L., Reynolds, T., Simpson, D., Barker, G., el-Rishi, H. Site formation processes in caves: the Holocene sediments of the Haua Fteah, Cyrenaica, Libya, Journal of Archaeological Science (2010), doi: 10.1016/j.jas.2010.01.021 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT Site formation processes in caves: the Holocene sediments of the Haua Fteah, Cyrenaica, Libya Chris Hunt1, John Davison1, Robyn Inglis2, Lucy Farr3, Tim Reynolds4, David Simpson1, Graeme Barker 2, Hwedi el-Rishi5 School of Geography, Archaeology, and Palaeoe olog , Quee s U i e sit of Belfast, UK Department of Archaeology, University of Cambridge, UK McDonald Institute for Archaeological Research, University of Cambridge, UK Faculty of Continuing Education, Birkbeck College London, UK Department of Geography, Garyounis University, Benghazi, Libya RI PT 1. 2. 3. 4. 5. D M AN U SC Abstract Caves have yielded some of the most globally important archaeological sequences, but often their interpretation has suffered from assumptions about cave sedimentary processes. Caves contain distinctive sedimentary environments: this has major implications for the understanding of contained archaeological materials. This paper describes and analyses the Holocene sediments in the Haua Fteah, a sequence regarded as essentially continuous by the original excavator. 50 years after it was first excavated the Haua s Epipalaeolithic to post-Classical chronological range and rich finds make it still the key Holocene archaeological site in North Africa. The reassessment shows, however, that the sequence is strongly discontinuous and this has major implications for the reinterpretation of the site, as the highly-resolved archaeological record is thus likely to reflect a series of brief occupations, rather than continuous human activity. As with many caves, the sedimentary record in the Haua Fteah is an extremely sensitive indicator of environments and processes in the wider landscape. Secure understanding of sedimentary process, from analysis of the highly individual records found in caves, is essential for full understanding of their contained archaeology. TE Keywords Cave archaeology, North Africa, Libya, Holocene, Facies Analysis, Radiocarbon Dating AC C EP Introduction Caves are often key archaeological sites because people used them over extended periods and because caves are known to preserve long and often critical archaeological and environmental records (e.g. Bar-Joseph & Meignen 2008; Bertran et al. 2008; Barker et al. 2007; Chase et al. 2009; Farrand, 2001; Hovers 2009; Hunt et al. 2007b; Laville et al. 1980; Mc Burney 1967). Yet caves are not simple repositories – they are dynamic sedimentary environments with their own complex taphonomic processes which shape the nature of the archaeological record (Hunt & Gale 1986). Understanding of taphonomic processes – the processes of site formation – are critical to the evaluation of the archaeological record from caves. Previous generations of cave archaeologists had only very limited understanding of sedimentary and taphonomic processes in caves, and this limited understanding of process had significant repercussions in their interpretation of the cave record. Typical early archaeological models of cave sedimentation often involve the concept of the build-up of a stead d izzle of sediment (e.g. Harrisson 1959a,b; McBurney 1967), resulting in archaeologists regarding cave deposits as containing continuous archaeological records. Modern re-evaluation of this early work, utilising sedimentological and stratigraphic studies of cave sediments (e.g. Woodward & Goldberg, 2001; Gilbertson et al. 2005; Bertran et al. 2008; Ghinassi et al., 2009), and taphonomic evaluation (e.g. Hunt & Rushworth 2004), is in 2 ACCEPTED MANUSCRIPT RI PT some cases providing dramatic new results (e.g. Woodward & Goldberg, 2001; Gilbertson et al. 2005; Barker et al. 2007; Bertran et al., 2008) diametrically at odds with the earlier interpretation. Caves contain highly complex sequences, laid down by a wide variety of processes (Gillieson, 1996; Gilbertson et al., 2005; Bertran et al., 2008; Ghinassi et al. 2009). Sedimentation rates vary by many orders of magnitude, both spatially and chronologically (Bailey & Galanidou, 2009). Deposition is often discontinuous and sometimes catastrophic. Erosion and recycling of materials occurs frequently. Yet in spite of the difficulty and complexity of cave records, it is clear that they are as significant archaeologically as ever. It has become increasingly evident, however, that the archaeological record from any cave is only as good as the understanding of the local depositional environment (Farrand, 2001; Butzer, 2008). M AN U SC In this paper we re-evaluate using a facies-analysis approach (Walker, 1992; Ghinassi et al., 2009) the Holocene depositional sequence in the Haua Fteah, Cyrenaica, Libya, as an example of how the re-evaluation of the cave sediments affects the understanding of the archaeological record. Here, the original interpretation (McBurney, 1967) was of a continually-deposited archaeological sequence, with sudden dramatic changes in the archaeology being thought to reflect major cultural events. This interpretation resulted from the interpretation of the sedimentary record as reflecting the more-or-less constant accumulation of the deposits, and thus that the archaeological sequence was virtually complete. Our re-evaluation shows a rather different pattern, which has major implications for our understanding of the nature of the contained archaeological record. AC C EP TE D Materials and methods The Haua Fteah, in the Gebel Akhdar, Cyrenaica, Northeast Libya (Fig. 1, 2), has remained one of the foundations of North African archaeology since the seminal publication of McBurney (1967). The Haua is technically a tafoni (a granular disintegration feature) on the side of a doline (a depression caused by solution and/or collapse common in limestone areas). The Haua Fteah has small relict phreatic features in places in its roof and small vadose conduits in its roof and floor which are active under very high rainfall, suggesting that the tafoni formed in an area already subject to cave formation. McBurney (1967) excavated to a depth of 14 m in the fill of this enormous feature during the 1950s, and discovered a highly stratified record which he suggested might reach back to the later part of the last interglacial period, perhaps 80000 years ago. M Bu e s e a atio s p odu ed what is still regarded as the longest sequence of human activity in North Africa. The site – and the prehistoric archaeology of the region - has been largely untouched by research since this ti e, a d M Bu e s findings are now in urgent need of revision. [Figs 1, 2 about here] The Cyrenaican Prehistory Project (CPP) is, therefore, carrying out a programme of archaeological fieldwork and science-based archaeology (Barker et al. 2007, 2008), with the aim of reassessi g M Bu e s o k a d i teg ati g it ith ode a haeologi al paradigms. It is clear from initial work (Barker et al. 2007, 2008, 2009, Simpson, Hunt 2009) that the Haua Fteah contains a high-resolution record, where changing environments and patterns of human activity can be discerned. It is also sho i g that a of M Bu e s findings require re-evaluation in the context of modern techniques and understandings. In this account, we conside the sedi e ts e posed i M Bu e s Uppe T e h , hich is a ~2 m deep, ~10 m by ~10 m excavation unit. This contained a Holocene sequence beginning ith the Li o-Capsia Mesolithi his Layers X, IX) dated using some pioneering 3 ACCEPTED MANUSCRIPT radiocarbon determinations by McBurney to c.8000-5000 BC, followed by the Neolithic (his Layers VIII-VI, 5000-2700 BC), and then the Graeco-Roman (his Layer III). The upper part of the trench (his Layers II, I) was referred to recent centuries. M Bu e s (1967) conception of the sedimentary processes operating seems to have been of the then-p e aili g d izzle of sediments model, with aeolian dust-fall augmented by various quantities of roof collapse. He therefore argued that the archaeological sequence in the cave was essentially complete. M AN U SC RI PT At the end of the 1955 season, the excavation was backfilled. Subsequently, local informants tell us that the Libyan authorities dumped gravel in the cave to provide a hardstanding for vehicles. During a high rainfall event around 1997-1998, considerable surface runoff entered the cave f o the doli e outside. Du i g this e e t … ud a d ate a i to the a e a d I fea ed that I ould e d o ed… (Mr Salim Hussein, pers. comm. 2007). The same informant described the traditional use of the cave as a place to manage and pen goats. Until recent years, the cave was used as part of a transhumant landuse pattern, with goat herds pastured at low altitude near the coast at the Haua during the winter, then moving inland to higher altitudes in the Gebel Akhdar for the spring flush of grazing and the summer. The build-up of ticks and other pests necessitated the burning of the accumulated guano on the cave floor at the end of the occupation season every one or two years. AC C EP TE D In April 2007 we removed the a kfill of the Uppe T e h to establish the robustness of the original faces and their suitability for renewed investigation. All of M Bu e s su i i g section pins and labels were recorded and the original faces were cleaned back substantially to expose fresh sediment. Detailed running sections (Figs. 3-4) were constructed on the western (East-facing) and southern (North-facing) faces and charcoal samples were taken from these sections for 14C dating. All radiocarbon dates were on well-preserved, angular charcoal roundwood pieces selected for their good preservation and lack of rounding as least likely to be recycled and without appreciable old carbon effects. Unpublished preliminary work on the sedimentary organic matter in the Holocene deposits by Simpson and Hunt shows that there is no appreciable movement of organic material through the sediments as the sedimentary organic matter profile is highly stratified and coherent. Hard water effects in this limestone region were avoided by selection of identified woody terrestrial taxa. Radiocarbon dates were by Accelerator Mass Spectrometry at the Oxford Laboratory for Archaeology and the history of Art. They were calibrated using Calib 5.02 and are documented in detail in Barker et al. (2008) and in Table 1. Selected profiles were cleaned further back, then samples taken from the East-facing section for sediment and other analysis: the work on these samples other than sediment analysis will be reported elsewhere. In this paper, determinations of magnetic susceptibility and of organic carbon by loss on ignition follow Gale & Hoare (1992). In the 2008 season we again cleaned the sections for investigation of the Holocene sediments in the Haua Fteah to establish the major depositional processes and postdeposition history, with a programme of intensive recording of sediment characteristics, sedimentary structures and pedogenic features, using drawing, photography, clast roundness and hand-texturing in the field. Krumbein roundness follows Krumbein (1941), using the procedure of Gale & Hoare (1992), and is expressed as minimum range value (mode) maximum range value, since means should not be calculated for ordinal data. The cave deposits The cave deposits are heterogeneous and analysis of the sections showed that they occur in unconformity-bounded units. Major changes in lithofacies (sedimentary characteristics 4 ACCEPTED MANUSCRIPT recording depositional environments) between and within beds are distinguishable. Lithofacies for sedimentary units in the two exposed faces (the East-facing and North-facing sections of the Upper Trench) are set out and interpreted below. Their occurrence is summarised in Tables 2 and 3, with their locations on the sections shown in Figs 3 and 4. Typical lithofacies are shown in Figs. 5-13, and sedimentary properties are shown graphically in Figs. 14 and 15. This lithofacies analysis omits pedogenesis and pedogenic features in the sediments, which will be dealt with elsewhere. RI PT [Figs 3-4 and Tables 1 and 2 about here] M AN U SC Diamicts, poorly mixed (D1) Diamict, unsorted, silty sand to gravel matrix, with clasts mostly subrounded (Krumbein roundness for Context 147 is 0.2(0.7)0.7: n=9) and randomly orientated. Also contains clods of calcreted riverine sediments, lumps of charred material, ashy lenses, dispersed carbonised and uncarbonised goat guano (Fig. 5). This layer is made ground: it in-fills the top of the McBurney trench above the primary fill. Local informants suggest that this layer was introduced into the cave and bulldozed into position during the 1980s a d it s te tu al qualities suggest derivation from local fan gravels. [Fig. 5 about here] AC C [Fig 6 about here] EP TE D Diamicts (D) Matrix to clast-supported diamicts, which consist of angular to subangular cobbles to small boulders in a silty matrix. Clasts mostly subangular (Krumbein roundness for Context 153 is 0.2(0.4)0.7: n=37; for Context 126 is 0.1(0.3)0.6: n=38) Clast li es lie at the ase of the deposit and also line scour-like features (Fig. 6). Clast long-axis imbrication is weak and predominantly horizontal, north-south orientated or slightly southward-dipping. The diamicts usually infill existing topography and wedge out southward. The matrix- and clastsuppo ted, poo l so ted atu e of these deposits, the asal sto e li e a d the sto e li es at the base of the scour-like features, together with the horizontal long axis orientation of clasts is consistent with these being mudflow deposits. Comparison with the field soils on the doline floor outside the cave, which are characterised by being predominantly silty with similar subangular clasts (Krumbein roundness 0.1(0.3)0.5: n=17), suggests that this is the sediment source. Clast-supported cobbles and boulder gravels (Gc) and Openwork cobbles and boulder gravels (Goi) Clast-supported and openwork gravel of cobble to small boulder size with angularsubangular clasts disorganised or showing a weak horizontal fabric which is north-south orientated (Fig. 7). Partly infilling some voids in what was originally openwork gravel are laminated silts, clayey silts and ash layers. This layer contains abundant carbonised goat guano. This layer is intimately related to the underlying diamicts - some large clasts in this layer are partly embedded in the diamict. The horizontal imbrication of some clasts in this openwork deposit is perhaps suggestive of deposition from rather turbid water flows, or watery mud-flows, probably at the end of the event which laid down the underlying diamict. Some of the silty void-fills probably relate to the settling of fines after the gravel was deposited. Later, other openwork voids were in-filled with waterlain ash and silts. 5 ACCEPTED MANUSCRIPT [Fig. 7 about here] RI PT Gravels (G) Clast to matrix-supported gravel, with subangular clasts in a silty-sand matrix. Clasts mostly subangular to angular (Krumbein roundness for Context 125 is 0.1(0.3)0.7: n=20, for Context 152 is 0.1(0.2)0.4: n=14) The horizontal long-axis imbrication of the clasts is strong. The sedi e t odies o up s ou s hi h a e V -shaped in cross-section, incised into older deposits (Fig. 8). The clast-to matrix-supported gravels with clear horizontal clast long-axis imbrication is suggestive of rapid deposition from relatively turbid flowing water. The morphology and sediment characteristics of these deposits are consistent with them being a waterlain gully-fill. [Fig. 8 about here] M AN U SC Silts (Si/Sil) Silts, sometimes with some fine sands, occasional small stones. The deposits are massive (Si) or indistinctly to distinctly southward-dipping cross- and plane-laminated (Sil). The clasts are subangular to angular (Krumbein roundness for Context 135 is 0.1(0.2)0.4: n=35; for context 136 is 0.1(0.3)0.4: n=10; for Context 132 is 0.1(0.2)0.2: n=6) The layers usually infill existing topography and wedge out southward. The generally well-sorted nature of these deposits and their southward dips is consistent with them being waterlain from shallow surface wash and rills originating on the doline floor outside the cave. [Fig. 9 about here] [Fig. 10 about here] TE D Laminated silts and clays (SiCl) Clayey silts, plane-laminated, with the laminations typically being silt-clay fining upward couplets (Fig. 10). The silt-clay fining upward couplets are consistent with these deposits being settling-out deposits from quiet water AC C EP Openwork breccia (Bo) Openwork breccia, with very angular clasts (Fig 11). Clast roundness for context 173 is 0.1 (n=25). The general depositional dip and clast long axes dip northward. In cross-section the sediment bodies are lenticular. The openwork texture and clast morphology of these deposits is consistent with them being debris-avalanche deposits resulting from rockfall. Breccia (B) Clast to matrix-supported breccia (Fig. 11). The matrix is silty, with some ash in places. The clasts are angular (Krumbein roundness for Context 144 is 0.1(0.2)0.4: n=25), with long axes conforming to the depositional dip. There is some size-sorting of clasts. Depositional dips are generally northward. The angular clasts and imbrication of these deposits are consistent them being debris-avalanche deposit resulting from rockfall, with a minor admixture of silty or ashy substrate which became entrained in the debris-avalanche. Ashy Breccia (Ba) Clast-supported to matrix supported breccia (Fig. 11). The matrix is mostly silt and/or ash and it contains artefacts, bone, shell and charcoal. The clasts are angular (Krumbein roundness for Context 131 is 0.1(0.1)0.5: n=24; for Context 161 is 0.1(0.1)0.4: n=21; for Context 172 is 0.1(0.1)0.4: n=10; for Context 167 is 0.1(0.2)0.4: n=20; for Context 168 is 6 ACCEPTED MANUSCRIPT 0.1(0.2)0.4: n=10) Clast long-axis orientation and depositional dip is generally northward. The sediment bodies are lenticular in cross-section, wedging out northward. The poorlysorted nature of these deposits, the general angularity of the clasts, the morphology of the sediment bodies all suggest that these are debris-avalanche deposits. The high content of anthropogenic material may be because debris avalanches incorporated this material from the substrate, or that it was anthropogenic material which collapsed to produce the debrisavalanches. RI PT [Fig. 11 about here] SC Tabular-bedded ash (At) Ash, silty and with occasional stones, containing artefacts, shell, and bone. The deposits are crudely and indistinctly tabular-bedded and usually very compact. Clast roundness is quite variable, probably reflecting a variety of parent materials: (Krumbein roundness for Context 133 is 0.1(0.1)0.8: n=11) The nature of these deposits, with their high anthropogenic content and compaction, suggests that they are occupation horizons. M AN U [Fig 12 about here] D Tabular-bedded ash and goat guano (Atg) Ash, stony ash and carbonised goat pellets, containing occasional to frequent artefacts, bone and shell. The layers typically are tabular in form, with a sharp upper contact and a g adatio al lo e o ta t, ofte ith fi ge s of ash i filli g s all u o s (Fig. 5). Clasts are angular (Krumbein roundness for Context 175 is 0.1(0.1)0.3: n=23). Discussion with local informants, together with the characteristics of the deposits - which are extremely rich in goat guano - suggest that these la e s a e sta le- u i g ho izo s, he e the floo of the cave was either deliberately or purposefully fired to eliminate ticks and other pests. EP TE Laminated ash (Al) Laminated or cross-bedded silty ash, containing occasional artefacts, bone and shell (Fig. 13). The layers infill depressions in the underlying topography and load-casts and waterescape structures are sometimes present. The clear laminations in these deposits are consistent with the accumulation of ash in quiet water following surface wash. [Fig. 13 about here] AC C Synthesis: lithofacies assemblages in the Holocene sediments of the Haua Fteah A number of major sedimentary facies assemblages are present in the Holocene deposits of the Haua. The lower part of the sequence, f o M Bu e s ) layers X-IV, can be assigned to two major facies assemblages, and the upper part to two further faciesassemblages. Lithofacies-assemblage 1 In the lower part of the northern half of the East-facing section, the dominant facies is silty and massive (structureless) or more-or-less laminated or cross-laminated (lithofacies Si, Si/Cl). The depositional dips in layers allocated to this facies are consistent with derivation from the doline floor, to the north. It may be suggested that these deposits result from shallow wash and rill-flow and the dominantly silt-grade sediments are most probably derived from ultimately aeolian sources, with comparatively little pedogenic alteration of the material prior to redeposition in its present location. The Krumbein roundness (Fig. 14) 7 ACCEPTED MANUSCRIPT is consistent with the clasts in these layers losing some of their angularity through transport in a flowing medium. SC RI PT Lithofacies-assemblage 2 In the lower parts of the north-facing section and the southern half of the east-facing section, the predominant facies association consists of more-or-less ashy breccias (lithofacies B, Bo and Ba) and occupation deposits (lithofacies At). Clasts in these breccias are characteristically very angular (Fig. 14) and have A-axis dips towards the north. The nonclastic component typically includes a high proportion of cultural debris. These can be assigned to rockfall (when predominantly limestone breccia: lithofacies B, Bo) and debrisavalanche processes (when containing significant ash and cultural material: lithofacies Ba), most probably in a dry state. It is hypothesised that the ashy breccias reflect cultural material spilling forward from an occupation area usually located on a slightly higher area of the cave floor, to the south of the a e, i to a su p a ea he e the M Bu e t e h is located. D M AN U Lithofacies-assemblage 3 This facies assemblage is distributed throughout the upper parts of the North-facing section and the East-facing section. This facies-assemblage, consisting of reddish clayey silts (lithofacies Si), diamicts (lithofacies D) and gravels (lithofacies G), results from mudflows and surface wash of field soils from the doline. This is inferred from the strong reddish-brown colours, sedimentary structures and clast morphology. Water-escape and loading structures suggest that the underlying sediments were in a wet state when some of these diamicts arrived. One major mudflow (contexts 126 and 153, correlating with M Bu e s La e III most probably arrived shortly before the building of the Graeco-Roman structure, but the major wash event which frightened our informant seems to have left little discernable trace in the sedimentary record. EP TE Lithofacies-assemblage 4 This facies assemblage is distributed throughout the upper parts of the North-facing section and the East-facing section and interdigitates with the third facies-assemblage. It consists of ashy deposits containing abundant carbonised goat pellets, derived from stabling, mostly of goats, and the burning of the resulting guano (lithofacies Atg and Al). AC C Facies change in the Haua and it’s causes The ajo sedi e ta ha ge elo M Bu e s la e III, f o the fi st t o of these fa iesassociations to the second pair, points to significant changes both inside and outside the cave. Inside the cave, the dominant process changes from debris–avalanche of occupation debris, to deposition of sediments resulting from goat-penning and stable-burning. It is probable that occupation by a residential group who kept their domestic animals elsewhere to the cave gave way at this point to shepherding communities predominantly using the cave for animal-keeping. There is also a change in the nature of the materials entering the cave at this time and the processes concerned. Small-scale wash of sediment derived from aeolian sources characterises the externally-originating layers in the earlier Holocene, before M Bu e s u it III. F o Da id “i pso s unpublished pollen analysis, the earlier Holocene vegetation of the area round the Haua was extremely open and of a rather arid aspect. This would chime with evidence for aeolian sedimentation outside the cave and a lack of pedogenic alteration of the materials entering the cave. This scenario is replaced in Historic ti es M Bu e s Unit III) by wash and mudflow of doline soils, most probably resulting from relocation of ploughsoil under extreme rainfall events. It is noteworthy that the major mudflow (contexts 126, 153) happened very shortly after the change in land-use outside the 8 ACCEPTED MANUSCRIPT RI PT cave and before the Graeco-Roman structure was built. It presumably reflects the relocation of easily-eroded stored sediment from the doline at a time when there was no vegetation cover, maybe as a result of ploughing, and once this had occurred, the quantity of sediment available for movement into the cave was substantially less. The rather uniform sedimentary properties throughout the Holocene (Fig. 14) reflect the rather similar, ultimately aeolian sources from which the Haua sediments have derived, but the switch between facies-associations 1 and 2 in the Libyco-Capsian and Neolithic layers and faciesassociations 3 and 4 in the Historic layers is visible as a general decline in magnetic susceptibility. This decline most probably reflects changes in magnetic mineralogy. as lowtemperature smouldering of goat guano during the Historic period generally higher temperatures reached by domestic fires during the Neolithic and Libyco-Capsian. M AN U SC The broad coincidence between these changes and the start of the Greek occupation in Cyrenaica around 621 BC may suggest that there were changes in residential location and land use associated with the incoming of the Greeks. The community previously resident in the cave was relocated and it would appear that the present-day pattern of cultivation of the doline and goat-herding in the cave started around this time. There is no trace of habitation activity in the sediments relating to the Graeco-Roman sanctuary. The vegetation on the Cyrenaican littoral in the Graeco-Libyan period consisted of forests with some cleared and grazed or cultivated areas (Hunt et al. 2002) but it was not as degraded as it became during the Roman occupation, or subsequently during the Little Ice age (Barker et al. 2008), but this increasing degradation in the wider landscape does not seem to have resulted in major sediment accumulation in the Haua Fteah. TE D Re-evaluation of the McBurney model of sedimentation This re-examination of the Holocene deposits in the Haua Fteah is only the first stage of the complete re-evaluation of this important depositional sequence. It provides a framework for detailed micromorphological, geochemical, palaeobotanical and palynological studies, and for the re-evaluation of the archaeology. AC C EP The new dating evidence and the facies analysis can be used to suggest that a re-evaluation of M Bu e s odel of o ti uous sedi e tatio is o e due. La e s see , on the whole, to have accumulated extremely rapidly, with significant hiatuses between depositional episodes (Table 3). In the Holocene layers underlying the 1980s made ground there are probably 24 significant depositional events, some of which probably took no more than a few minutes to occur, and few of which lasted more than a few years. This is particularly the case for the mudflow, debris-avalanche and rockfall deposits, which would have accumulated virtually instantaneously, but it is also likely to be the case with the waterlain layers, which most probably accumulated in single wet seasons. This pattern of episodic deposition has been reported widely in other recent high-resolution studies of cave sedimentation (e.g. Otte et al., 2003; Gilbertson et al., 2005; Lundberg & McFarlane, 2007; Pickering et al. 2007; Bertran et al., 2008; Ghinassi et al. 2009). There are also a few instances of significant recycling – the date for layer X of 12761-12131 BC is far older than would be expected from the other dates associated with this layer both from McBurney (1967) and from Barker et al. (in press), who have calibrated dates of 8806-8615 BC and 9294-8951 BC from lower in layer X deposits. This is a phenomenon increasingly to be expected from caves and many other types of archaeological sites as technological advance enable the taking of smaller radiocarbon samples and more cost-effective protocols lead to increases of the average number of dates per site. 9 ACCEPTED MANUSCRIPT RI PT A new model of sedimentation for the Haua Fteah would thus be of long periods of hiatus separated by extremely rapid and often catastrophic sedimentation events incorporating, on occasion, materials from older levels. This change in the interpretation of the sequence do not negate its archaeological significance, but they do require a re-evaluation of their contained archaeology, which would now perhaps be reinterpreted as highly-resolved s apshots of a haeologi al e e ts athe tha a o ti uous se ue e. A further corollary is that (as the excavated locality is the topographic low in the cave and thus where virtually all material might be expected to accumulate) the history of human activity in the Haua Fteah was extremely discontinuous, with short phases of activity separated by many years of abandonment. This may be the reason that McBurney (1967) documents a series of highly distinctive artefact assemblages in these horizons - conceivably people were gradually modifying their technology during the episodes of non-deposition, but the artefacts we see come only from short, well-separated periods of activity in the cave and thus are likely to be typologically distinct. M AN U SC An alternative (but by no means completely mutually exclusive) viewpoint (e.g. Bailey & Galanidou, 2009) would hold that these assemblages are time-averaged palimpsests derived from the accumulation of many years human activity on a stable surface. At present, the unabraded and pristine nature of the lithics and the extremely good preservation of bone, molluscs and charred plant debris recovered by the deep section cleaning and the first of the new excavations in the cave inclines us against this hypothesis, since in time-averaged palimpsests there is a strong tendency for degradation of archaeological materials by passing feet, weathering and animal activity. D These viewpoints will be tested further over the coming years by highly-controlled testexcavations. It is clear that the full significance of the archaeological record at the Haua will not be apparent until tests of the coherence of layers, such as lithic refitting and fluorescence microscopy of pollen grains (Hunt et al. 2007a) have been carried out. TE [Table 3 about here] AC C EP Conclusions: cave geoarchaeology and the quality of the archaeological record The example of the Haua Fteah confirms many indications from the recent literature. In many cases, caves do not contain simple sedimentary systems, but extremely complex and variable ones. The deposition of sediment in caves is intimately related to the processes occurring at any given point in time and space: these processes will vary in time, and across the cave floor, with distinctive facies relating to the geography and geometry of the cave, the location and types of inputs of sediment and the transport mechanisms involved. People may further complicate these patterns by their activities, and the location and type of human activity is governed by cultural issues, the geography and geometry of the system, the stability and dryness of the cave floor and the cave microclimate. Also of considerable significance are periods of non-deposition, where successions of cultural activities may occur over considerable periods. Caves, during times when they are accumulating archaeologically-important deposits, are intimately related to the outside world, and the processes which transport sediment into and deposit it within caves are often, but not always, manifestations of processes which are operating outside. Sediment supply also is likely to vary through time, so that even if processes are constant, the nature and rate of sediment accumulation may vary. Caves thus mirror the outside world in many ways - that is part of their value to the geoarchaeologist and to the wider archaeological community. And - as with the outside world - transport, depositional and erosional processes will vary with 10 ACCEPTED MANUSCRIPT time and with changes in the outside climate and environment. Also significant will be changes in the morphology of the cave itself. RI PT The geoarchaeologist wishing to interpret cave sequences has a series of key issues to address. First and possibly most important is the identification of the sedimentary mechanisms involved in the formation of the cave record. The variety of processes involved is considerable, but facies analysis offers a powerful tool for their identification. Second, the recognition of changing sedimentation rate and of periods of non-deposition is critical to the i te p etatio of the se ue e a d it s o tai ed a haeolog . This has eso a e i the recognition of - on one hand - instantaneous s apshots o lage state a d - on the other, palimpsests and time-averaged deposits. Without secure geoarchaeological understanding, the cave archaeological record is meaningless. TE D M AN U SC Acknowledgements We would like to particularly acknowledge the permission and support of the Department of Antiquities of Libya to undertake the project; the financial support of the Society for Libyan Studies, the Leakey Foundation, and the University of Cambridge, for the fieldwork; and the significant help in kind provided by the Department of Antiquities in Tripoli and Cyrene. The personal support of the President of the Department of Antiquities Dr Giuma Anaag and the Controllers of Antiquities for Cyrene Achmed Sabar and Abdulgader al Mulzeni has been vital to the success of the project. Further invaluable support for the project was provided by Paul Bennett (former Chair of the Society for Libyan Studies), Mustapha Turjman (Department of Antiquities, Tripoli), Ibreike Quinhe (Inspector of Antiquities, Apollonia), Mohammed Twati and Fathallah Khalifa (Department of Archaeology, Omar Mukhtar University al-Baeida), and Ahmed Buzaian (Department of Archaeology, Gar Yunis University, Benghazi). We thank the many members of the Cyrenaica Prehistory Project for support in the field and discussion. An early version of this paper was read at the World Archaeological Congress, Dublin, 2008. EP References Bailey, G., Galanidou, N. 2009. Caves, palimpsests and dwelling spaces: examples from the Upper Palaeolithic of south-east Europe, World Archaeology,41:2 ,215 — 241. AC C Bailey, G. N., Woodward, J. 1997. The Klithi deposits: sedimentology, stratigraphy and chronology. In Klithi: Palaeolithic Settlement and Quaternary Landscapes in Northwest Greece, Vol. 1, Excavation and Intra-site Analysis at Klithi (ed. G. N. Bailey). Cambridge: McDonald Institute for Archaeological Research, pp. 61–94. Barker, G., Hunt, C., and Reynolds, T., Brooks, I., el-Rishi, H. 2007. The Haua Fteah, Cyrenaica (northeast Libya): renewed investigations of the cave and its landscape, 2007. Libyan Studies 38: 93-114. Barker, G., Basell, L., Brooks, I., Burn, L., Cartwright, C., Cole, F., Davison, J., Farr, L., Grün, R., Hamilton, R., Hunt, C., Inglis, R., Jacobs, Z., Leitch, V., Morales, J., Morley, I., Morley, M., Pawley, S., Pryor, A. Reynolds, T., el-Rishi, H., Roberts, R., Simpson, D., Stimpson, C., Touati, M., van der Veen, M. 2008. The Cyrenaican Prehistory Project 2008: the second season of 11 ACCEPTED MANUSCRIPT investigations of the Haua Fteah cave and its landscape, and further results from the initial (2007) fieldwork. Libyan Studies 39, 175-221. RI PT Barker, G., Antoniadou, A., Barton, H., Brooks, I., Candy, I., Drake, N., Farr, L., Hunt, C., Abdulhamid Ibrahim, A., Inglis, R., Jones, S., Morales, J., Morley, I., Mutri, J., Rabett, R., Reynolds, T., Simpson, D., Twati, M. & White, K. 2009 The Cyrenaican Prehistory Project 2009: the third season of investigations of the Haua Fteah cave and its landscape, and further results from the 2007-2008 fieldwork. Libyan Studies 40, 55-94. Bar-Yosef, O. Meignen, L. (eds.) 2008. Kebara Cave, Mt Carmel, Israel. The Middle and Upper Paleolithic Archaeology. Part I. Harvard, Harvard University Press SC Be t a , P., Ca e , L., La goh , ‘., Le ée, L., d E i o, F. . Continental palaeoenvironments during MIS 2 and 3 in southwestern France: the La Ferrassie rockshelter record. Quaternary Science Reviews 27, 2048–2063. M AN U Butzer, K. W. 2008. Challenges for a cross-disciplinary geoarchaeology: The intersection between environmental history and geomorphology. Geomorphology 101, 402–411. Chase, P.G., Debenath, A, Dibble, H.L.. McPherron, S.P. 2009. The Cave of Fontechevade. Recent Excavations and Their Paleoanthropological Implications. Cambridge, Cambridge University Press Farrand, R. 2001. Sediments and stratigraphy in rock shelters and caves: a personal perspective on principles and pragmatics. Geoarchaeology 15.5: 537-557. D Gale, S. J., Hoare, P. G. 1992 Quaternary Sediments: Petrographic methods for the study of unlithified rocks. London, Belhaven. EP TE Ghinassi, M., Colonese, A. C., Di Giuseppe, Z., Govoni, L., Lo Vetro, D., Malavasi, G., Martini, F., Ricciardi, S., Sala, B. 2009 The Late Pleistocene clastic deposits in the Romito Cave, southern Italy: a proxy record of environmental changes and human presence. Journal of Quaternary Science 24, 4, 383-389. AC C Gilbertson, D. Bird, M. Hunt, C. McLaren, S. Mani Banda, R. Pyatt, B. Rose, J., Stephens, M. 2005. Past Human activity and geomorphological change in a guano-rich tropical cave mouth: initial interpretations of the Late Quaternary succession in the Great Cave of Niah, Sarawak. Asian Perspectives 44, 1, 16-41. Gillieson, D. S. 1996. Caves, development, processes and management. Oxford, Blackwell, 324 pp. Harrisson, T. 1959. New archaeological and ethnological results from the Niah Caves. Man 5: 108. Hovers, E. 2009. The Lithic Assemblages of Qafzeh Cave. Oxford, Oxford University Press. Hunt, C. O., El-Rishi, H. A., & Hassan, A. T. 2002 Reconnaissance investigation of the palynology of Holocene wadi deposits in Cyrenaica, Libya. Libyan Studies 33, 1-7 12 ACCEPTED MANUSCRIPT Hunt, C. O., Rushworth, G., Dykes, A. P. 2007a UV-fluorescence microscopy and the coherence of pollen assemblages in environmental archaeology and Quaternary geology Journal of Archaeological Science 34, 4, 562-571. Hunt, C. O., Gilbertson, D. D., Rushworth, G. 2007b Modern humans in Sarawak, Malaysian Borneo, during Oxygen Isotope Stage 3: palaeoenvironmental evidence from the Great Cave of Niah. Journal of Archaeological Science 34, 1953-1969. RI PT Hunt, C. O., Rushworth, G. 2005 Airfall sedimentation and pollen taphonomy in the West mouth of the Great Cave, Niah. Journal of Archaeological Science 32, 465-473. Laville, H., Rigaud, J.P., Sackett, J., 1980. Rock Shelters of the Perigord: Geological Stratigraphy and Archaeological Succession. Academic Press, New York. 371pp. SC Lundberg, J., McFarlane, D.A. 2007. Pleistocene depositional history in a periglacial terrane: A 500 k.y. record from Kents Cavern, Devon, United Kingdom. Geosphere 3, 4, 199-219. M AN U Krumbein, W. C. 1941 Measurement and geological significance of shape and roundness of sedimentary particles. Journal of Sedimentary Petrology 11, 64-72. McBurney, C.B.M. 1967. The Haua Fteah in Cyrenaica. Cambridge: Cambridge University Press. D Otte, M., López Bayón, I., Noiret, P., Bar-Yosef, O., Yalçı ka a I., Ka tal, M. Jean-Marc Léotard, J.-M., Pettitt, P. 2003 Sedimentary Deposition Rates and Carbon-14: the Epi-paleolithic Sequence of Öküzini Cave (Southwest Turkey). Journal of Archaeological Science 30, 325–341. TE Pickering, R., Hancox P. J., Lee-Thorp, J. A., Grün, R., Mortimer, G. E., McCulloch, M., Berger, L. R. 2007. Stratigraphy, U-Th chronology, and paleoenvironments at Gladysvale Cave: insights into the climatic control of South African hominin-bearing cave deposits. Journal of Human Evolution 53, 602-619. AC C EP Simpson, D.J., Hunt, C. O. 2009 Scoping the past human environment: a case study of pollen taphonomy at the Haua Fteah, Cyrenaica, Libya. Archaeological Review from Cambridge 24, 2, 27-46. Walker RG. 1992. Facies, facies models and modern stratigraphic concepts. In Facies Models: Response to Sea Level Changes, Walker RG, James NP (eds). Geological Association of Ca ada: “t Joh s; –14. 13 ACCEPTED MANUSCRIPT CAPTIONS TO THE FIGURES Fig. 1. Map of the central Mediterranean showing the location of the Haua Fteah. RI PT Fig. 2. The Haua Fteah before re-excavation in 2006, showing also the cultivated field in the doline in front of the cave. The site of the McBurney excavation is visible as a large craterlike depression in the cave fill. Fig. 3. The East-facing Section in the re-excavated Upper Trench in the Haua Fteah, showing the McBurney layers, the contexts identified by the CPP, and the locations of the new radiocarbon dates obtained by the CPP. Modified from Barker et al. (2009). SC Fig. 4. The North-facing Section in the re-excavated Upper Trench in the Haua Fteah, showing the McBurney layers and the contexts identified by the CPP. M AN U Fig. 5. Detail of the East-facing section, showing D1 diamict of context 147 overlying Atg ash with carbonised goat pellets of context 135 and then Si clayey sitls of context 138. Divisions on scale are 0.1 m. Fig. 6. Detail of the North-facing section, showing D diamict of context 153 with large clasts floati g i a silt at i a d ith a asal li e of lasts. To the left of the photog aph, a scour-like features is lined with clasts. Divisions on scale are 0.1 m. D Fig. 7. Detail of the East-facing section showing context 125 - Gc clast-supported gravel (darker matrix) passing up into Goi openwork gravel subsequently infilled with waterlain sandy silt (lighter matrix). Divisions on scale are 0.1 m. TE Fig. 8. Detail of the junction between the East-facing and North-facing sections showing a gully-fill with basal G imbricated clast-supported gravel of context 152. Divisions on scale are 0.1 m. EP Fig. 9. Detail of the East-facing section showing context 135 with massive and crosslaminated silts Sim and Sil. Divisions on scale are 0.1 m. AC C Fig. 10. Detail of the East-facing section showing weakly laminated very clayey silts SiCl of context 138. Divisions on scale are 0.01 m. Fig. 11. Detail of the North-facing section showing strong brown silty diamict D of context 153 (at top of picture) overlying dark grey ashy breccia Ba of context 161, overlying light brown openwork breccia Bo of context 173 (in middle of picture) overlying light brown clast to matrix-supported breccia B (in lower part of picture). Divisions on scale are 0.1 m. Fig. 12. Detail of East-facing section, showing shelly tabular silty ash At occupation horizon, overlain by cross-bedded and massive silts Si of context 132. Divisions on scale are 0.1 m. Fig. 13. Detail of East-facing section, showing crudely-laminated silty ash Al in context 140, with carbonised goat pellets in the lower part of the picture. Divisions on scale are 0.1 m. Figure 14. Sedimentary properties of the sample profile of Holocene deposits in the Eastfacing section of the Upper Trench in the Haua Fteah. 14 ACCEPTED MANUSCRIPT AC C EP TE D M AN U SC RI PT Figure 15. Krumbein roundness histograms for contexts in the Holocene sediments of the Haua Fteah. Context 173 (lithofacies Bo) is excluded from the diagram since a sample of 25 clasts all had a Krumbein roundness of 0.1. Table ACCEPTED MANUSCRIPT Table 1. Details of CPP and McBurney (1967) Holocene radiocarbon dates from the Haua Fteah, modified after Barker et al. (2009). McBurney (1967) states his dated aterial o l as har oal a d his re ords do ot e a le attri utio to CPP o te ts. Sample and species CPP Context McBurney Layer Phase Lab no. Radiocarbon age Cal. BC/AD 190 (Juniperus sp.) 243 (Zygophyllum sp.) 126 129 III VI Historic Neolithic OxA-18710 OxA-18675 1688±25 5759±28 243 (Zygophyllum sp.) 129 VI Neolithic OxA-18676 5462±30 243 (Zygophyllum sp.) 129 VI Neolithic OxA-18794 5521±32 Charcoal na VI top Neolithic NPL-40 5800±108 Charcoal na VI Neolithic NPL-41 4860±97 209 (Rhus cf. tripartita) 131 VIII Neolithic OxA-18673 244 (Suaeda sp.) 131 VIII Neolithic OxA-18667 238 (Ephedra sp.) 132 VIII Neolithic OxA-18674 380 (Rhus cf. tripartita) 132 VIII Neolithic OxA-19028 6413±32 Charcoal na VIII Neolithic W98 6800±350 Charcoal na VIII Neolithic NPL-42 6370±103 254 (Juniperus sp.) 136 X Capsian OxA-18678 12360±50 317 (Juniperus sp.) 176 X Capsian OxA-19184 9425±40 319 (Juniperus sp.) 176 X Capsian OxA-19158 9740±45 Charcoal na X Capsian GrN-3167 8400±150 Charcoal na X Capsian W89 7300±300 Charcoal na X Capsian Grn-3541 7000±110 259-416 AD 4692-4534 BC 4356-4259 BC 4451-4331 BC 4932-4375 BC 3936-3375 BC 5877-5729 BC 5208-4949 BC 5533-5376 BC 5471-5326 BC 6400-5001 BC 5526-5065 BC 1276212130 BC 8806-8615 BC 9294-8951 BC 7729-7061 BC 6822-5570 BC 6062-5673 BC RI PT SC 6115±31 6505±33 M AN U D TE EP AC C 6917±31 (2σ) Table ACCEPTED MANUSCRIPT Table 2. Lithofacies identified in the Holocene sediments of the Haua Fteah. G Matrix to clast-supported diamicton consisting of subangular cobbles to small boulders in a silty matrix. Weak clast long-axis imbrication is predominantly north-south. In the upper part, laminated silts, clayey silts and ash layers between clasts. Layers usually infill existing topography and wedge out southward. Diamict, unsorted, clasts mostly rounded and randomly orientated, clods of calcreted riverine sediments, lumps of charred material, ashy lenses, dispersed carbonised and uncarbonised goat guano. Clast to matrix-suported gravel, clasts subrounded-subangular, longaxis imbrication follows depositional dip, silty matrix. Sheets of gravel or lenticular sediment bodies occupying gullies. Silts, sometimes with fine sands, massive or indistinctly to distinctly southward-dipping cross- and planelaminated, occasional small stones. Layers usually infill existing topography and wedge out southward. Silts, sometimes clayey, well laminated. Openwork breccia, clasts very angular, general depositional dip and clast long axes dipping northward. Clast long-axis orientation and depositional dip generally northward. Lenticular sediment bodies in cross-section. Breccia, clast to matrix-supported, silty, some ash in places. Clasts angular, with long axes conforming to depositional dip. Some sizesorting. Depositional dips generally northward. Breccia, clast-supported to matrix supported, matrix mostly silt and/or ash and containing artefacts, bone, Mudflow deposits originating on the doline floor outside the cave during heavy rainfall events. Infiltration deposits between clasts in the upper part of the units. 126, 142, 146, 149, 153 Made ground: layer of fill introduced into the cave and bulldozed into position during the 1980s. 147, 148 TE B Ba EP Bo AC C Sil RI PT Context D Si Interpretation SC D1 Description M AN U Lithofacies code D Gravel deposited from sediment-rich flows at end of mudflow, waterlain gully-fill gravel. 125, 152 (lower part) Wash and rill deposits originating on the doline floor outside the cave during heavy rainfall events. 132, 135, 136, 138, 150, 154, 155, 156, 400 Settling-out deposits from quiet water Debris-avalanche deposits resulting from rockfall. 128, 134 Debris-avalanche deposit resulting from rockfall. Minor admixture of cultural material. 144, 169, 170, 176, 177 Occupation debris, rockfall and aeolian silt relocated in small debris 129, 130, 131, 143, 151, 159, 173 ACCEPTED MANUSCRIPT 133, 402 Laminated or cross-bedded silty ash, containing occasional artefacts, bone and shell. Accumulation of ash in quiet water. RI PT Occupation debris SC Burning of goat guano layers. M AN U D TE Al 161-168, 172, 401 EP Atg avalanches. AC C At shell and charcoal. Clasts angular to slightly subangular, clast long-axis orientation and depositional dip generally northward. Lenticular sediment bodies in cross-section, wedging out northward in long section. Ash, silty and with occasional stones, containing artefacts, shell, bone, crudely and indistinctly tabularbedded. Ash and carbonised goat pellets, occasional to frequent artefacts, bone and shell. 135, 137, 145, 158, 159, 175, 176 17, 127, 139, 140, 141, 146, 152 (upper part), 157 Table ACCEPTED MANUSCRIPT Table 3. Stratigraphic relationships, depositional environment and CPP dates from layers in the Haua Fteah Date D1 Made ground 1980s AD Si Si G Atg 1955-1980s AD 4 138, 146, 156, 157 149 Si 3 Gully-fill ashy silt Gully-fill ashy silt Gully-fill gravel Goat penning and stable-burning Wash and rill, silt with ash Al 4 145, 158, 159 125, 144, 143, 145, 139 126, 142, 153 128, 400 129 Atg 131, 172 132 402 162, 163 I–II Historic Sil, G 3 IV Historic D 3 Ba VII Neolithic 2 EP VI Neolithic TE Sil D III Historic SC Post 1955 Faciesassemblage Ash accumulating in quiet water Goat penning and stable-burning M AN U Lithofacies code RI PT Depositional environment 130, 161 McBurney layer AC C Layer code no. 148, 147 154 155 152 137 Settling out deposits from quiet water, silts and ash, infilling openwork gravels Mudflow originating in doline floor Settling out deposits from quiet water Debris avalanche relocating occupation debris and fallen rock Ba 2 Debris avalanche relocating occupation debris and fallen rock Debris avalanche relocating occupation debris and fallen rock Ba 2 Si 1 Wash and rill deposits originating in doline floor At Ba 2 Occupation layer Occupation debris relocated from the 259-416 AD 4692-4534 BC, 43564259 BC, 4450-4331 BC 5877-5729 BC 5208-4949 BC 5533-5376 BC 5471-5326 BC ACCEPTED MANUSCRIPT 167 (pars) VIII Neolithic Ba 2 133 134 IX LibycoCapsian At SiCL 2 1 Si 143, 168 Ba 144, 169, 170, 177, 176 135, 172 B 175 167 (pars) B Ba 2 M AN U 404 RI PT Ba SC 164, 166, 165 EP X LibycoCapsian TE D Si/Ba AC C 136 Ba/Si south by debris avalanche Occupation debris relocated from the south by debris avalanche Occupation debris relocated from the south by debris avalanche Occupation layer Settling out deposits from quiet water Wash and rill deposits originating in doline floor Occupation debris relocated from the south by debris avalanche Rockfall breccia 2 Occupation debris relocated from the south by debris avalanche, grading into wash deposits from the doline floor Rockfall breccia Occupation debris relocated from the south by debris avalanche Occupation debris relocated from the south by debris avalanche, grading into wash deposits from the doline floor 12760-12130 BC (probably recycled: see Table 1) Figure 1 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 2 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 3 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 4 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 5 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 6 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 7 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 8 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 9 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 10 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 11 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 12 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 13 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT Figure 14 Click here to download high resolution image AC C EP TE D M AN U SC RI PT ACCEPTED MANUSCRIPT MANUSCRIPT AC C EP TE D M AN U SC RI PT Figure 15 Click here to download high resolution image ACCEPTED