ArticlePDF Available

Abstract and Figures

We use stone tool refitting to assess palimpsest formation and stratigraphic integrity in the basal units of the Gruta da Oliveira archeo-stratigraphic sequence, layers 15–27, which TL and U-series dating places in late Marine Isotope Stage (MIS) 5 or early MIS 4. As in most karst contexts, the formation of this succession involved multiple and complex phenomena, including subsidence, bioturbation, carnivore activity and runoff as agents of potential post-depositional disturbance. During phases of stabilization, such as represented by layers 15, 21 and 22, the excavated area was inhabited and refits corroborate that post-depositional displacement is negligible. Layers 23–25 and 16–19 correspond to subdivisions that slice thick geological units primarily formed of material derived from the cave’s entrance via slope dynamics. Refit links are consistent with rapid fill-up of the interstitial spaces found in the Karren-like bedrock (for layers 23–25), or left between large boulders after major roof-collapse events (for layers 16–19). Layers 26 (the “Mousterian Cone”) and 27 are a “bottom-of-hourglass” deposit underlying the main sedimentary body; the refits show that this deposit consists of material derived from layers 15–25 that gravitated through fissures open in the sedimentary column above. Layer 20, at the interface between two major stratigraphic ensembles, requires additional analysis. Throughout, we found significant vertical dispersion along the contact between sedimentary fill and cave wall. Given these findings, a preliminary analysis of technological change across the studied sequence organized the lithic assemblages into five ensembles: layer 15; layers 16–19; layer 20; layers 21–22; layers 23–25. The lower ensembles show higher percentages of flint and of the Levallois method. Uniquely at the site, the two upper ensembles feature bifaces and cleavers.
Content may be subject to copyright.
Assessing site formation and assemblage
integrity through stone tool refitting at Gruta
da Oliveira (Almonda karst system, Torres
Novas, Portugal): A Middle Paleolithic case
Marianne Deschamps
*, João Zilhão
1UNIARQ – Centro de Arqueologia da Universidade de Lisboa, Faculdade de Letras de Lisboa,
Universidade de Lisboa, Alameda da Universidade, Lisboa, Portugal, 2CNRS UMR 5608 TRACES,
´de Toulouse Jean Jaurès, Toulouse, France, 3Institucio
´Catalana de Recerca i Estudis Avanc¸ats
(ICREA), Barcelona, Spain, 4Universitat de Barcelona, SERP (Seminari d’Estudis i Recerques
Prehistòriques, SGR2014-00108), Departament d’Història i Arqueologia, Facultat de Geografia i Història,
Barcelona, Spain
We use stone tool refitting to assess palimpsest formation and stratigraphic integrity in the
basal units of the Gruta da Oliveira archeo-stratigraphic sequence, layers 15–27, which TL
and U-series dating places in late Marine Isotope Stage (MIS) 5 or early MIS 4. As in most
karst contexts, the formation of this succession involved multiple and complex phenomena,
including subsidence, bioturbation, carnivore activity and runoff as agents of potential post-
depositional disturbance. During phases of stabilization, such as represented by layers 15,
21 and 22, the excavated area was inhabited and refits corroborate that post-depositional
displacement is negligible. Layers 23–25 and 16–19 correspond to subdivisions that slice
thick geological units primarily formed of material derived from the cave’s entrance via slope
dynamics. Refit links are consistent with rapid fill-up of the interstitial spaces found in the
Karren-like bedrock (for layers 23–25), or left between large boulders after major roof-col-
lapse events (for layers 16–19). Layers 26 (the “Mousterian Cone”) and 27 are a “bottom-of-
hourglass” deposit underlying the main sedimentary body; the refits show that this deposit
consists of material derived from layers 15–25 that gravitated through fissures open in the
sedimentary column above. Layer 20, at the interface between two major stratigraphic
ensembles, requires additional analysis. Throughout, we found significant vertical disper-
sion along the contact between sedimentary fill and cave wall. Given these findings, a pre-
liminary analysis of technological change across the studied sequence organized the lithic
assemblages into five ensembles: layer 15; layers 16–19; layer 20; layers 21–22; layers 23–
25. The lower ensembles show higher percentages of flint and of the Levallois method.
Uniquely at the site, the two upper ensembles feature bifaces and cleavers.
PLOS ONE | February 16, 2018 1 / 34
Citation: Deschamps M, Zilhão J (2018) Assessing
site formation and assemblage integrity through
stone tool refitting at Gruta da Oliveira (Almonda
karst system, Torres Novas, Portugal): A Middle
Paleolithic case study. PLoS ONE 13(2): e0192423.
Editor: Michael D. Petraglia, Max Planck Institute
for the Science of Human History, GERMANY
Received: October 31, 2017
Accepted: January 23, 2018
Published: February 16, 2018
Copyright: ©2018 Deschamps, Zilha
˜o. This is an
open access article distributed under the terms of
the Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
Funding: Over the years, the excavation and study
of the Gruta da Oliveira site and finds were funded,
logistically supported, or aided by by grants-in-kind
received from the following entities: Fyssen
Foundation post-doc fellowship http://www. (to Dr. Marianne
Deschamps). TRACES - UMR 5608 of CNRS http:// (to Dr. Marianne Deschamps).
Even though stone tool refitting goes back to the end of the 19th century [12], it is only since
the 1960’s that it has been widely used to reconstruct past technological systems and assess
assemblage integrity and site formation process. Building on the “chaîne opératoire” concept
derived by Leroi-Gourhan [3] from Mauss’s [45] ethnographic “chaîne technique” [6], stone
tool refitting was systematically applied for the first time at Pincevent (Seine et Marne, France),
a site whose pristine conservation of occupation surfaces represented an ideal scenario for the
application of the technique [7]. Core reduction sequences and the economy of debitage prod-
ucts were reconstructed in detail, and the distribution of refitted elements was used to make
inferences about the contemporaneity of different loci, the spatial organization of activities, or
the social structure of the hunter-gatherer bands that had discarded the analyzed remains [8
10]. The Pincevent example has since been emulated at sites presenting a similar degree of
preservation [1112], while the analysis of vertical distributions has been used to assess the
impact of palimpsest formation and post-depositional disturbance at both open-air and cave
or rock-shelter stratified sites [1322].
With regards to the Middle Paleolithic, most refitting studies have so far concerned sites in
the open-air [2331], even though a few examples of its application in rock-shelter contexts
also exist [16,3239]. Here, we add to this limited corpus a case study from Gruta da Oliveira
(Almonda karst system, Torres Novas, Portugal). This site contains a ca.13 m-thick strati-
graphic succession rich in Middle Paleolithic stone tools and bone remains, including human
ones, and is dated by a combination of different methods to the interval between ca.35 and
ca.105 ka (thousands of years) [4043]. Our study targeted the basal part of the sequence, lay-
ers 15–27. As implied by the minimum age imposed by the TL dating of overlying layer 14 to
78±8 ka [43], these basal units formed during early MIS (Marine Isotope Stage) 4 or late MIS
Based on the observations made during excavation, coupled with the micromorphological
analysis of stratigraphic thin sections, the geoarcheological study of the deposit concluded that
the accumulation of layers 15–27 had proceeded through roof and wall collapse episodes alter-
nating with phases of stabilization [40]. The presence throughout of subsidence-deformation
features and of roots and animal burrows, indicated, however, that a degree of post-deposi-
tional disturbance was to be expected, while long-distance vertical displacement along voids
between the cave walls and the sedimentary column represented an additional cause of poten-
tial contamination (by material derived from higher up in the sequence).
A refitting study carried out on the lithic assemblages from the upper part of the sequence
(layers 8–14) had shown the impact of these processes to be limited [44], but the issue
remained to be assessed for the lower part. A systematic intra- and inter-level refitting project
was therefore designed as a prerequisite to the techno-economic analysis of the collections
from layers 15–27. The aim was to define and constitute the analytical units best suited for the
study of the stone tool assemblages in all possible dimensions, e.g., raw-material provenience
and selection, technology, typology, debitage economy, use-wear, and spatial distribution.
The questions of wider methodological interest that we addressed are as follows:
Do the archeo-stratigraphic units defined at the time of excavation constitute valid units for
the organization of the stone tool collections into behaviorally meaningful assemblages, or
do those units need to be grouped, or further subdivided, and, if so, how?
Which parts of the excavation trench were most affected by post-depositional disturbance,
how significant was such an impact, and, if significant, how do we separate the reliable from
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 2 / 34
Fundac¸ão para a Ciência e Tecnologia project
index.phtml.en (to Dr. João Zilhão). Associac¸ão
´gica do Algarve (to Dr. João Zilhão).
ˆmara Municipal de Torres Novas (to Dr. João
Zilhão). Crivarque S.A. (to Dr. João Zilhão).
Instituto Português de Arqueologia projects
– Fa
´brica de Papel do Almonda (to Dr. João
Zilhão). STEA — Sociedade Torrejana de
Espaleologia e Arqueologia (to Dr. João Zilhão) The
funders had no role in study design, data collection
and analysis, decision to publish, or preparation of
the manuscript. The specific roles of these authors
are articulated in the ‘author contributions’ section.
Competing interests: The authors have no
association with Crivarque S.A. There are no
patents, products in development, or marketed
products to declare. This does not alter the
authors’ adherence to all PLOS ONE policies on
sharing data and materials.
the unreliable units of provenience, and how is such a separation affected by the variation
observed along the sequence in the topography of the encasing bedrock?
Which of the meaningful/reliable assemblages produced by the application of the tapho-
nomic filters are amenable to paleo-ethnographic analysis of activity (those with limited
palimpsest formation and attendant good preservation of the distributions’ original spatial
structure), which can only lend themselves to analysis of long-term processes (those with sig-
nificant palimpsest formation and attendant loss of the distributions’ original spatial struc-
ture), and which conclusions of chrono-stratigraphic significance can be drawn from the
changes observed through time in the composition of the assemblages?
For most of its duration, the Middle Paleolithic of Europe falls outside the limits of applica-
bility of radiocarbon, and alternative dating methods are of much lower precision. Thus, the
period’s chronological structure remains largely based on the patterns of change in technology
and typology observed across reference sequences, and in the correlation of such sequences
using lithic assemblage composition patterns coupled with paleoenvironmental information
[4548]. The potential impact of functional variability on such patterns has long been the
object of much debate (e.g. [4950]), but the extent to which the units of analysis used in the
different schemes that have been proposed are indeed stratigraphically valid has often been
overlooked. Yet, in stratified sites of the cave and rock-shelter type, post-depositional distur-
bance and palimpsest formation are bound to have impacted assemblage composition—and
the more so in periglacial contexts. We believe, therefore, that the approach we used to tackle
these problems at Gruta da Oliveira is of wider interest and, hence, justifies dedicated publica-
tion of our methodology and results.
Materials and methods
The site
The spring of River Almonda is a karst outflow located at the base of a ca.70 m-high cliff face,
part of the NE-SW, ca.40 km-long fault escarpment that separates Portuguese Estremadura’s
Central Limestone Massif form the Cenozoic basin of the Tagus. Due to the tectonically-
induced rising of the massif relative to the basin, the underground sections of the river under-
went a process of downward migration in the karst network that left behind, at higher eleva-
tion, an interconnected, labyrinthine system of passages leading to fossil springs that once
functioned as cave entrances. With time, these entrances filled-up with sediment, collapsed,
and became covered by slope deposits. A program of systematic speleo-archeological survey of
the system initiated in the late 1980s has since discovered, re-exposed and excavated a few [51
52] (Fig 1). Gruta da Oliveira, whose entrance lies at an elevation of ca.115 m, was the first
such collapsed cave to be found—in 1989—and investigated—over some 20 field seasons end-
ing in 2012 [4043,5359].
Going in from the re-opened entrance, the Gruta da Oliveira begins as a now unroofed
porch area, the “Exterior,” leading to a short, meandering passage developed along the fault
through which the Almonda progressively incised its underground course. This “Access Corri-
dor” is, on average, ca.10 m-high and 3 m-wide. Some 10 m inward from the collapsed
entrance, the Access Corridor branches into three sections that, even though located at differ-
ent elevations, remain linked by narrow fissures and joints: at the bottom, the “Passage of the
Column” (Galeria da Coluna), which communicates the cave with the “Passage of the Sieve”
(Galeria do Crivo) and the depths of the endokarst beyond; higher-up, to the left, the “Septem-
ber 27 (27-S) Chamber” (Sala 27 de Setembro), a low and wide space (ca.4×4 m and, on aver-
age, ca.2.5 m-high) developed along a stratification joint perpendicular to the main fault; and,
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 3 / 34
at the top, to the right, the “Side Passage” (Divertículo), a narrow conduct that leads to the Pas-
sage of the Sieve through a different underground path (Fig 2).
The minimal unit used in the excavation of the Gruta da Oliveira was the “spit” or décapage.
Spits are slices of the deposit differentiated to constrain the horizontal and vertical position of
non-piece-plotted or sieve-collected finds. Spit thickness varied between 15 to 20 cm, when
going through levels primarily made of large e
´boulis, and 5 cm, when dealing with finer sedi-
ments with potentially good preservation of spatial distributions. In all cases, the delimitation
of a spit’s basal surface followed the general dip of the stratification and respected observed
stratigraphic boundaries whenever they imposed a reduction of the originally targeted thick-
ness. From top to bottom of the deposit, spits were numbered sequentially, per square meter
unit of the grid or per unit of open-area excavation, from 1 to n, preceded by the letter A (e.g.,
spit A1, spit A70, etc.).
At the time of excavation, spits were grouped into “layers,” defined on the basis of features
that could be followed across significant horizontal extents and were suggestive of change in
Fig 1. The Almonda karst system. 1. Geographical location and raw-material sources (reprinted from [59] under a CC BY license, with permission from Journal of
Lithic Studies, original copyright 2016). 2. Drone view from NE over the fault escarpment separating Portuguese Estremadura’s Central Limestone Massif from the
Cenozoic basin of the Tagus; the rectangle indicates the position of the network of underground passages associated with the spring of River Almonda. 3. Drone
view from the South over the cliff face above the spring showing the position of the currently known cave entrances that have been archeologically excavated
(Galeria da Cisterna and Lapa dos Coelhos—Later Prehistory, Neolithic, and Upper Paleolithic; Gruta da Oliveira—Middle Paleolithic; Grutada Aroeira—
Acheulean; Gruta do Pinheiro—Middle or Lower Paleolithic hyena den). 4. Schematic cross-section of the Almonda cliff face along the main fault, with indication
of the archeological localities currently known in the staircase of fossil passages above the spring.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 4 / 34
Fig 2. Gruta da Oliveira: Site layout and excavation views. 1. Plan and excavation grid; wall contours are shown at
the elevation of layer 10; the colored grid units indicate the position of the initial findspot, the “Mousterian Cone,” a
deposit underneath the sedimentary column that, at the time of discovery, blocked the Passage of the Column. 2. View
from the Access Corridor onto the two branches at the back end of the cave, the 27-S Chamber and the Side Passage;
the archeo-stratigraphic units (layers) differentiated in these sections of the site are indicated. 3. View over the Access
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 5 / 34
the dynamics of the accumulation, such as: base-of-boulder planes defining the cave floor
extant at the time of roof-collapse events; flowstone or calcite-incrusted surfaces; increases in
the clay content of the matrix; and associated color changes. As some of these “layer” units
span the three main sections of the cave plan (Access Corridor, 27-S Chamber, and Side Pas-
sage), the different sequences excavated in each section could be laterally correlated and inte-
grated into a single stratigraphic succession scheme [40].
The Gruta da Oliveira layers are the equivalent of the Geoarcheological Field Unit (GFU) as
theoretically defined in the context of the excavation of the Lagar Velho rock-shelter (Leiria,
Portugal) [60]: a three-dimensional body composed of sediments, of either natural or cultural
origin, that differ from the surrounding ones in patterned manner. Once bedrock was reached
and the excavation completed, the comprehensive view of the succession thusly obtained led,
however, to the grouping of the excavation’s layers/GFUs into ensembles that come closer to
true stratigraphic units in the geological sense. As currently recognized, based on Zilhão et al.
(2013), such ensembles can be described, from bottom to top, as follows (Figs 24).
Passage of the Column and Mousterian Cone (layers 26–27). Basal units of the succes-
sion, the initial discovery of which, in 1989, triggered the site’s subsequent excavation. The
Mousterian Cone was an e
´boulis cone with a silty-sand matrix that blocked the speleological
progression towards the outside for cavers approaching the cliff face from the Passage of the
Sieve. Fossilized remains of a Pleistocene fauna associated with Middle Paleolithic stone tools
with no signs of transport-damage were observed on the surface, and its quarter cubic-meter
test excavation, carried out in 1990, yielded some 200 stone tools and 150 bones and bone frag-
ments, 30% of which burnt [52]. This cone was interpreted as a “bottom-of-hourglass” deposit
whose matrix and archeological components had filtered down through fissures linking it to
an archeological cave fill inferred to exist above. The inference led to the search for the cave’s
entrance, eventually found, re-opened and excavated—the Gruta da Oliveira. When reached
from the top, in 2012, the Mousterian Cone was found to be the equivalent of layer 26 of the
overall succession, from which, however, it was largely disconnected. At that time, sediment
was also observed in fissures at the base of the Mousterian Cone area; forming the lateral
equivalent of the Passage of the Column’s fill, that deposit was excavated as layer 27.
Access Corridor Lower Ensemble (layers 23–25). Large rock masses stuck sub-vertically
between wall and wall, supporting the overlying stratification, and roofing the empty space at
the end of the Passage of the Column that enabled the original detection of the Mousterian
Cone (Fig 2, panel 3). A degraded stalagmitic crust separated the upper 25 cm of this ensemble
(layer 23) from the ca.65 cm (layer 24) that remained until bedrock was reached in column N
of the grid. Layer 25 was mostly the e
´boulis, ca.80-90 cm-thick, that, in columns O-P of the
grid, filled, down to the elevation of the Mousterian Cone’s surface, the narrow space commu-
nicating the Access Corridor with the Passage of the Column. The fine matrix between boul-
ders and blocks contained a low-density scatter of bones and artefacts distributed across the
ca.1.5 m thickness of the deposit but mostly coming from column P, which was significantly
affected by roots and featured significant voids along the contact between sediment and cave
wall (Fig 4). Burnt flints and burnt bone were found in layer 23 but, given the evidence for
post-depositional disturbance, were considered at the time of excavation as probably related to
the hearth features found in layer 22 above.
Corridor from its back end taken during the excavation of spit A67 (layer 21); note the voids along the contact between
sedimentary column and cave walls, the hearth feature in the middle of the trench, and, bottom left, the direct
connection with the Mousterian Cone and the Passage of the Column beyond.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 6 / 34
Fig 3. Gruta da Oliveira: Topographic cross-section along the intersection between columns P and R of the grid.
Note the complete colmatage of the cave by the sedimentary column. The Mousterian Cone (layer 26) is a “bottom-of-
hourglass” deposit filtered down through voids in the basal e
´boulis and along the cave walls that also fills the Karren-
like bedrock of the Passage of the Column below (layer 27). Elevations are in cm below site datum.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 7 / 34
Access Corridor Middle Ensemble (layers 20–22). Silty loam, ca.65 cm-thick, formed
during a period of relative structural stabilization that allowed use of this part of the site for
habitational purposes—as documented by a circular hearth feature with a diameter of ca.1.5 m
Fig 4. Gruta da Oliveira: The reference cross-section (N-P15>14) for the Access Corridor’s lower levels. The dotted lines indicate the
boundaries between the three geo-stratigraphic ensembles recognized here. The subsidence-deformed dark bands at the base of layer 22
represent the edges of large hearth features extending outward beyond the trench wall. Elevations are in cm below site datum.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 8 / 34
whose archeological content and enveloping sediment were differentiated as layer 21 (Fig 2,
panel 3). The edges of two other, subsidence-deformed features of similar size that extend out-
wards, beyond the extant wall of the trench, were excavated in underlying layer 22 (Fig 4). The
base of these features defined the boundary with the upper unit of the Lower Ensemble, layer
23. Layer 20 is marked by the resumption of roof-collapse events, including the fall of a mas-
sive, 2 m-long, 50 cm-wide, 60 cm-thick erosional blade that deformed the layer 21 hearth.
Access Corridor Upper Ensemble (layers 15–19). Silty loam with a variable clay compo-
nent, ca.1.3 m-thick, filling the space between large boulders. These boulders signal a phase of
major structural destabilization with detachment of multi-ton rock masses whose bases
defined the boundaries between layers 18 and 19 and between layers 19 and 20—the latter also
marked by a carbonate crust, almost continuous but degraded by post-depositional phosphati-
zation. In the reference section for the Access Corridor, this ensemble is capped by a stalag-
mitic crust, and discontinuous carbonate incrustations, sometimes forming thin but discrete
crusts, were also observed at the corresponding elevation (top of layer 15) in the grid units
leading to the 27-S Chamber. Across most of this ensemble, the characteristics of the sediment
suggest that its archeological content relates to in-wash, through low-energy run-off, or over-
land flow, of material abandoned in the context of occupations taking place a few meters out-
ward, in the cave’s porch—even though the largely horizontal disposition of layer 15 and its
higher clay content suggest accumulation in a phase of relative stability and a decrease in the
contribution of derived material. The upper units of the Access Corridor Upper Ensemble spill
onto the 27-S Chamber: layer 17 abuts the steeply inclined rock face bridging it with the Access
Corridor; layer 16 extended laterally to fill the space between the protruding ridges of the 27-S
Chamber’s Karren-like bedrock, thusly forming the relatively regular floor upon which layer
15 came to lie.
Basal Cave Interior (layers 13–14). The Access Corridor Upper Ensemble is capped by a
>20-ton, ca.3 m-thick chunk of the roof fallen on grid units O-P/12-15 and obstructing com-
munication with the outside. The silty loam filling-up the space behind this large rock mass
was subdivided at the time of excavation into two layers with largely horizontal upper and
lower boundaries that form the main body of the 27-S Chamber deposit: layers 13 and 14. The
latter is characterized by a finer texture, with a higher clay content and a more intense reddish
color. As shown by the presence of a hearth feature and corroborated by stone tool refitting
[44], the archeological finds made in these two layers relate to in situ human occupation of the
27-S Chamber and adjacent areas of the Access Corridor.
Middle Cave Interior (layers 9–12). As the 27-S Chamber filled-up, the cave’s interior
space became more constrained and mostly functioned as a hyena den, with the remains of
limited human occupation being found in the Side Passage and adjacent areas of the Access
Corridor. The boundary with layer 13, marked by a stone plane and an increase in the sand
and clast components of the deposit, is clear. Detailed descriptions of these units can be found
in [40].
Upper Cave Interior (layers 7–8). Units burying the O-P/12-15 rock mass, eventually
bringing about the colmatage of the cave’s interior space and limiting human occupation to
the Exterior section of the site. Layer 7 is archeologically sterile, while the artefacts and faunal
remains retrieved in layer 8 largely correspond to an inward, gravity-displaced tail of the origi-
nal distribution.
Colmatage breccia. Large boulders, e
´boulis and matrix forming a heavily cemented brec-
cia that clogged the cave’s porch and the space behind as far inward as the Side Passage. The
lower boundary of this ensemble is given by a well-developed, very well cemented carbonate
flowstone, with clear laminar structure and a maximum thickness of 13 cm. This flowstone
sealed the archeological sequence across the Access Corridor and the Side Passage, and, inward
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 9 / 34
of the colmatage breccia’s rim, underlay massive stalagmitic columns that blocked the access
to the 27-S Chamber and largely obstructed the Side Passage’s communication with the deeper
reaches of the system.
Through the sequence, but especially in the Access Corridor, the “Swiss cheese” nature of
the karst network provided all along for the existence of interstitial, void spaces underneath
and to the sides of the sediment column. Therefore, subsidence adjustments, and the attendant
generation of cracks and fissures, must have been recurrent through the accumulation, espe-
cially along the contact with the walls but sometimes also within the sedimentary body itself.
In addition, structural layout activity was a permanent factor of instability, one that led to the
frequent collapse of large rock masses fallen from the roof, on one hand, and to the dislodge-
ment of huge chunks of the cave wall sliding along the fault’s plane, on the other. Finally,
cementation by carbonate precipitation induced significant lateral variation within individual
units, either layers or ensembles. In the Access Corridor, differential sediment induration was
most apparent in the Lower Ensemble and the Middle Ensemble. Across these units, the areas
adjacent to the wall in column N tended to be brecciated, while those adjacent to the opposite
wall in column P were looser and often bioturbated.
An important implication of the complex topography of the encasing bedrock is the stepped
nature of the sedimentary accumulation’s baseline. Because of this, the lowermost reaches of
the sedimentary fill of the 27-S Chamber and the Side Passage remained open for long periods.
Indeed, the interstitial spaces of their irregular, Karren-like floors often feature variably large,
cemented remnants of the fluvial sands and silts accumulated when the Almonda flowed out
through Gruta da Oliveira. These remnants are overlain by a stony cave earth accumulated
during an interval that we can only very loosely constrain: between the time when the subterra-
nean watercourse migrated downwards and the time when the accumulation of sediments in
the Access Corridor reached the plane defined in those adjacent, higher-up sections by the
ridges of their Karren-like bedrocks. Such intra-Karren cave earth fills are therefore susceptible
of containing remains related to the lower-down occupation of the Access Corridor discarded
through the accumulation of the latter’s Lower, Middle, Upper and Basal Cave Interior ensem-
bles. To account for this possibility, the lateral equivalents of layers 16 and 13 of the Access
Corridor have been differentiated as, respectively, layer 16bis in the 27-S Chamber, and layer
13bis in the Side Passage. Likewise, the deposit overlying layer 13 of the 27-S Chamber corre-
sponds to the open, bioturbated, inward tail of the Middle (layers 9–12) and Upper (layers
7–8) Cave Interior ensembles and, accordingly, has been differentiated as layer 12bis.
The research questions
The site formation process inferred from the stratigraphic observations made at the time of
excavation and from the geoarcheological analysis of the succession begged several questions
requiring a lithic taphonomy study. For instance, the preservation of fire features in layers 21
and 22 implied that the associated finds be largely in situ, in both the geological and the arche-
ological senses, and this was indeed corroborated by the successful refit of 18 pieces document-
ing the reduction of a Levallois flint core in layer 21, grid units O/15-16, (refit set R-2000; Fig
5, panel 5) [61]. The inference remained to be tested, however, for the other units of the Access
Corridor Middle Ensemble, layers 20 and 22.
Conversely, the mode of deposition of the Access Corridor Lower and Upper ensembles
implied that a significant proportion of their archeological content be derived, along the talus
of an inward-dipping slope, from occupations taking place farther out. Notwithstanding,
observations made at excavation showed that intra-level refits existed, and the presence of
articulated faunal remains (e.g., several elements of a lion’s paw at the interface between layers
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 10 / 34
19 and 20) supported that post-depositional disturbance be limited. Thus, these field observa-
tions remained consistent with two alternative modes of accumulation: (a) each of the individ-
ual layers differentiated in the Access Corridor’s Lower and Upper ensembles during
excavation could correspond to the re-deposition of a single lens made up of items originally
discarded in the context of occupations taking place in the site’s Exterior section; or (b) the
Fig 5. Gruta da Oliveira: Typical 3D dispersion patterns as exemplified by four refitting sets. 1. projection of the items in each of the color-coded refit set against
the vertical z-axis and the corrected horizontal axis (corrY, left; corrX, right); the position of the corrected horizontal axes used relative to the excavation grid is given
in the middle. 2–5. the refit sets in the projections. Each set is identified by a single color, used for the dots in the projections and the boxes bounding their photos. The
cloud of black dots in the background represents all of the site’s piece-plotted artefacts. The R- prefix denotes the identification number of individual refit sets.
Elevations are in cm below site datum.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 11 / 34
integrity of such original occupations could have been lost, or significantly compromised,
along the re-deposition path, in which case each of those individual excavation layers could
potentially contain items that, in the primary locus of deposition—the Exterior area of the site
—had been part of stratigraphically distinct lenses. Finding out which of these alternatives was
the correct one required a systematic investigation of potential inter-level links. For these
ensembles, a related question was whether the upper units of each, layers 23 and 15 (which
represented phases of stabilization at the end of sedimentary accumulation processes leading
to the complete burial of major roof-collapsed rock masses), could feature levels of assemblage
integrity akin to layer 21’s.
Given the relatively imprecise boundaries of the chronometric results available for layers
15–27, which remained consistent with deposition intervals ranging between a minimum of
7000 (94–101 ka) and a maximum of 45,000 (62–107 ka) years [43], these questions had ancil-
lary chrono-stratigraphic implications. For instance, cleavers and bifaces were found in the
Access Corridor Upper Ensemble (layers 15–19), but were entirely absent both above and
below [42]. Finding that each of those Upper Ensemble layers stood for a stratigraphic unit
sensu stricto could lend support to the notion that this part of the archeo-stratigraphic
sequence represented a significant amount of time, spanning several climate-driven deposi-
tional cycles; hence, its duration could have been closer to the maximum allowed by the chro-
nometric results and the cleaver-plus-biface episode could have been relatively long-lived.
Finding the opposite would in turn support a short duration for the archeo-stratigraphic
sequence as much as for that characteristic lithic industry.
Another question that needed to be addressed was the nature and genesis of the Passage of
the Column and Mousterian Cone infillings (layers 26–27). Previous interpretations oscillated
between the original view that they were largely formed of vertically displaced material [52]
and the hypothesis that they mostly represented material horizontally derived (washed in)
from occupations emplaced outward [42]—a combination of both processes being also con-
ceivable. Sorting out these alternatives required the inclusion of layers 26–27 in the overall
refitting study, and the more so because the back end of the sedimentary column above opened
to an empty space providing direct, unimpeded communication with the Mousterian Cone—
the narrow fault running obliquely along units P-S/18-19 of the grid (Fig 2, panel 3).
The approach
Our study was carried out in the premises of UNIARQ–Centro de Arqueologia da Universidade
de Lisboa (Alameda da Universidade, 1600–214 Lisboa, Portugal), the host institution of the
project to investigate the archeology of the Almonda karst system and where the Gruta da Oli-
veira finds are in storage. The project is carried out under excavation permits issued by the rel-
evant authorities of the Government of Portugal (over the years: IPPAR–Instituto Português do
Património Arquitectónico e Arqueológico; IPA–Instituto Português de Arqueologia; DGPC–
Direcção-Geral do Património Cultural; excavation director, João Zilhão). The ongoing study
of the finds is carried out under the 2015–2018 research project “ARQEVO–Arqueologia e Evo-
lução dos Primeiros Humanos na Fachada Atlântica da Península Ibérica” (principal investiga-
tor, João Zilhão), DGPC-approved under its 2015 call for Planos de Investigação Plurianual em
To investigate the issues above we focused on the quartzite component of the stone tool
assemblage. This choice was dictated by several reasons:
A significant proportion of the flints presents a whitish patina masking variation in color
and surface appearance and rendering the identification of individual blocks more difficult
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 12 / 34
than in the case of the quartzite items because these, when patinated, are only slightly so,
and, in addition, are highly variable in color, texture and cortex.
Based on the raw-material procurement study carried out for layer 14 [59], the main flint
varieties used come from at least 15 km away (even though closer-by, <5 km away sources
were also used), while quartzite is locally available within a radius of few hundreds of meters.
The probability that complete reduction sequences could be represented in the lithic assem-
blages was therefore significantly higher for quartzite than for flint.
Quartz is also locally available, and was widely used at Gruta da Oliveira. However, this raw-
material is less suited for refitting studies because of its homogeneity in appearance and fre-
quent failure to follow the principles of conchoidal breakage. In addition, the use of power
tools in the excavation of the indurated parts of the deposit entailed accidental breaks, the
recent surfaces of which are more difficult to identify with quartz than with flint or quartzite.
Excluding hammerstones and unmodified manuports, the total number of quartzite items
catalogued in layers 15–27 is 3541. Since the inclusion of chippage and small flakes would not
have been cost-effective, our study was restricted to cores, flakes, tools and thermoclasts >2
cm. In the field inventory, the count for these categories among piece-plotted items is 1718,
which, pending completion of the collections’ technological study, provides a good estimate of
the size of the sample targeted by our refitting work (Table 1). Once individually labelled, the
finds were laid out on large tables, separated by stratigraphic provenience and sorted according
to descriptive criteria (size of quartz grains, color, sub-cortical alteration). After an initial
phase of intra-level analysis, links with adjacent units were searched for; eventually, the search
was extended to non-adjacent units.
For the assessment of assemblage integrity and post-depositional disturbance, refit sets link-
ing items with more than one stratigraphic unit of provenience were assigned to a “layer of
original production” (LoP) (Table 2,S1 and S2 Tables; Fig 5). This assignment was based on
the three-dimensional distribution of the sets’ components and the following assumptions,
derived from field observations, the geoarcheological data, and the results of the refitting study
previously carried out for layers 8–14:
As a rule, post-depositional vertical displacement occurred downwards rather than upwards.
For any given unit, items retrieved in the syn-depositionally indurated areas of the Access
Corridor (broadly, columns N and O) are more likely to be in situ than items retrieved in the
looser, potentially bioturbated areas of the same unit (broadly, column P).
Items retrieved in burrowed or root-disturbed areas explicitly identified as such in the field
documents as well as items found adjacent to the cave wall (in column P) or to the steeply
inclined bedrock plane bridging the Access Corridor and the 27-S Chamber (i.e., depending
on stratigraphic depth, in rows 18–19 or 17–18) are most susceptible to represent long-dis-
tance displacement and, hence, contamination of a given unit’s assemblage by material
intruded from above.
In some cases, no LoP could be assigned, namely when the refitting set was composed of
only two items coming from either potentially disturbed areas or layers 26–27. More rarely,
assignment was impossible because the distribution spanned non-adjacent units and the direc-
tion of the displacement could not be ascertained.
The refit sets for which LoP’s could be proposed were then classified into categories based
on the nature of the links, defined according to the stratigraphic distance encompassed: intra,
when only items from a single layer were found in the set; adjacent, when items from an
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 13 / 34
immediately underlying or overlying layer were included; long, when the links extended
beyond the units adjacent to the assigned LoP. These categories were considered “normal” as
the distributions respected stratigraphy within site formation and excavation constraints: on
one hand, links with units adjacent to the assigned LoP are to be expected as a result of surface
Table 1. Refit success rate (quartzite). Variation in the percentage of refitted items (debitage, debris, and thermoclasts) per stratigraphic unit of provenience.
Layer Refitted (N) TotalPiece-plotted
N Refitted N Refitted
15 83 993 8% 421 20%
16 26 253 10% 148 18%
16bis 7 55 13% 23 30%
17 57 307 19% 165 35%
18 78 487 16% 234 33%
19 94 433 22% 215 44%
20 63 336 19% 209 30%
21 15 120 13% 58 26%
22 45 190 24% 109 41%
23 15 53 28% 36 42%
24 4 30 13% 16 25%
25 1 6 17% 2 50%
26 22 138 16% 46 48%
27 15 140 11% 36 42%
TOTAL 525 3541 15% 1718 31%
cobbles, hammerstones and other unmodified manuports excluded
 used as a proxy for the larger-sized material targeted by the refitting study
Table 2. Gruta da Oliveira typical refit sets. Classification and interpretation after the nature of stratigraphic links and rationale for layer of production (LoP)
Archeo-stratigraphic unit of provenience for items in the
refit set
Total LoP Nature of
15 16 16
17 18 19 20 21 22 23 24 25 26 27
1055 – – – 1 – 2 1 – – – – – – 4 20 Adjacent The layer 18 item was found in a bedrock fissure in N18; it was
labelled "layer 18" in the field based on elevation criteria. This
item predates the accumulation of layer 18 and is an instance at
lower elevation of the problems generated by the open nature of
the sediment filling interstitial spaces of the 27-S Chamber’s
Karren-like bedrock that underpin the differentiation of layer
16bis (see text).
1074 – – 1 – 1 – – – 1 – – 1 – 4 ? Anomalous The layer 17 item comes from square O17, close to the wall but in
a brecciated area, and represents the parsimonious LoP for this
set. However, the distribution’s scatter does not allow secure
inference and no LoP is assigned.
1133 19 1 1 1 – – – – – – – – – 22 15 Long The items from layers 16, 17 and 18 come from square P18 and
are downwardly displaced along the cave wall. The layer 15 items
all come from the 27-S Chamber. This refit set is a good example
of the processes affecting the stone tool assemblage (broad
stratigraphic integrity coupled with some vertical displacement
between adjacent levels and occasional long-distance links
provided for by wall effects and bioturbation).
2000 – – – – – 2 16 – – – – – – 18 21 Intra The layer 20 items are from the the interface with layer 21 and
reflect décapage error, not post-depositional displacement (see
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 14 / 34
dynamics, bioturbation, and excavation error; on the other hand, all links beyond adjacent
units in sets for which an LoP nonetheless could be assigned concerned items found within
pockets of localized post-depositional disturbance (e.g., burrows), along the encasing bedrock
in areas affected by wall effects (e.g., column P), in the Mousterian Cone/Passage of the Col-
umn, or in the fill of the 27-S Chamber’s basal Karren (layer 16bis). The refitting sets with
inter-layer links that could not be explained per these factors and for which no LoP could be
assigned were considered anomalous (“non-normal”).
The rationale for considering as “normal” the sets in the adjacent category of refit links
derives from the diffuse nature of most stratigraphic interfaces, compounded by the difficulty
encountered in indurated areas when trying to follow them. A case in point concerns the indi-
vidualization of layer 21. The plane defined by the black staining found at the base of spit A66,
denoting the upcoming emergence of a large hearth (Fig 2, panel 3), was defined as the lower
boundary of layer 20. Layer 21 was in turn defined as the 5–10 cm-thick slice of the deposit
excavated as spit A67 down to the base of that feature. In the periphery of the latter, especially
in column N, however, the induration of the sediment often requested the power tool-assisted
removal of chunks that sometimes were themselves as thick as, or thicker than A67 itself. Con-
sequently, for finds made at this elevation, a measure of error is expected to exist in the spit-to-
layer correlation and, hence, in the stratigraphic assignment made in the field (e.g., to layer 21
instead of layer 20, or to layer 22 instead of layer 21, and vice-versa). Refit links bridging these
layers are therefore likely to reflect such kinds of “excavation error” rather than post-deposi-
tional disturbance—which, given the very preservation of the layer 21 hearth feature, can only
have been minimal.
Likewise, limited exchange of items across the interface between two adjacent layers is to be
expected due to trampling, small-scale burrowing, and progradation (the latter given the posi-
tion of the Access Corridor trench, at the back end of the inward sloping connection with the
Exterior area). Such interface porosity introduces a measure of fuzziness in the definition of
assemblages, but is of no consequence for the overall typological and technological assessment
of the material retrieved within the thick sediment packages found above and below the
boundaries of defined stratigraphic units. The same logic applies to the cases when a refit set
links a given LoP with items displaced downward through wall effect, even if as far down as
the Mousterian Cone/Passage of the Column—the sets whose refit links fall in the long cate-
gory. Such connections document loss to erosion and the contamination of underlying units
of the sequence, but are of no consequence for the assessment of the stratigraphic integrity of
the displaced items’ layer of origin.
Bearing in mind these data and premises, we analyzed the stratigraphic integrity of individ-
ual units of the sequence in light of: (a) the projection of linked items along the axis of deposi-
tion, which is oblique to the axis of the grid and so required prior transformation of the (x,y)
field coordinates into corrected ones (corrX, corrY) (for the sake of simplicity, in these projec-
tions the material from layer 16bis has been considered together with layer 16); (b) the tabula-
tion of the proportion of sets with intra and adjacent refit links within individual LoP’s; and
(c) the tabulation of the same proportions for the ensembles of production (EoP) into which
the analysis of (a) and (b) led us to group the different archeo-stratigraphic units.
The refitted sets are 150, with the most complete comprising 22 objects. R-2000 is a flint refit,
the other 149 are all quartzite. The variation across the sequence in the frequency of the differ-
ent categories of refit sets is given in Tables 3and 4. In Table 3, the data are presented per indi-
vidual LoP. In Table 4, the same data are organized into the analytical ensembles judged to be
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 15 / 34
those best suited for the organization of the collections. With regards to the research questions
that we targeted, these data support several conclusions.
Genesis of the Mousterian Cone/Passage of the Column
We find that several long-distance refits link layers 26–27 with all other units of the overlying
sequence. In addition, only two intra-level refits were made within this ensemble, each com-
prised of no more than two items (Fig 6). This evidence shows that the stone tools found
herein do not form a coherent assemblage. They represent syn- or post-depositional, down-
ward displacement of items whose LoP spans the entire sequence targeted by our study,
explaining the high refit success rate obtained—measured against the totals for piece-plotted
items, among the highest in the sequence (Table 1).
These items’ displacement process may have consisted of (a) gradual, attritional, intermit-
tent “drizzling” of artefacts along the voids between the sediment column and the cave walls,
(b) mass displacements of chunks of the sedimentary fill in the context of such catastrophic
events as those underpinning the accumulation of the multi-ton boulders lying on the surface
of layers 20, 19, and 15, or (c) a combination of both. The two intra-level refits suggest the dis-
placement of finds with maintenance of original connections, which supports (b) or (c)—as
does the fact that, so far, no refit links with layers 8–14 exist. With present evidence, however,
the alternative cannot be resolved. The one conclusion that can be drawn for now is that layers
Table 3. The different types of refit sets. Distribution per layer of production (LoP).
LoP sets (N) Nature of stratigraphic links (N) Intra+adjacent Anomalous
Intra Adjacent Long Anomalous
15 28 16 3 9 68%
16 6 0 1 5 17%
17 10 4 2 4 60%
18 23 6 8 9 61%
19 25 8 11 6 76%
20 14 4 4 6 57%
21 5 3 1 1 80%
22 15 6 2 7 53%
23 2 – 2 0%
24 – – –
25 – – –
?22 2 5 0 15 32% 68%
TOTAL 150 49 37 49 15 57% 10%
Table 4. The different types of refit sets. Distribution per ensemble of production (EoP).
EoP sets (N) Nature of stratigraphic links (N) Intra+adjacent Anomalous
Intra Adjacent Long Anomalous
15 28 16 7 5 0 82%
16–19 74 42 16 16 0 78%
20 14 4 6 4 0 71%
21–22 21 11 7 3 0 86%
23–25 3 0 3 0 0 100%
?10 2 5 0 3 70% 30%
TOTAL 150 75 44 28 3 79% 2%
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 16 / 34
26–27 do not form a valid analytical unit for the study of the cave’s archeology and so their
content ought to be removed from considerations relating to diachronic change in the human
use of the site.
Assemblage integrity
In layers 23–25, lithic artefacts are few and, corroborating field observations, refits primarily
demonstrate links with layer 22, from where, at the time of excavation, the burnt, hearth-
related bones and flints retrieved in layer 23 were inferred to derive. When the horizontal dis-
tribution of the piece-plotted items from this layer is considered (Fig 7), the pattern is easy to
explain. Indeed, most come from areas that were subject to strong subsidence and attendant
deformation and disturbance effects: (a) the ca.1 m-wide band along the P-column wall of the
cave, which is also bioturbated (as clearly seen in the reference cross-section; Fig 4), or (b) the
back end of the trench—atop the coarse, often openwork e
´boulis forming layers 24–25 or
above the large void situated between the sedimentary column and the Mousterian Cone. The
few and scattered finds made in the middle of the trench and against the opposite wall come
from apparently undisturbed areas, and that is likely to be the prevalent condition in the unex-
cavated areas of the site outwards from the Access Corridor trench. Be it as it may, the stone
tool sample from layers 23–25 available to us mostly comes from an area of the trench where
the integrity of the deposit cannot be warranted. Therefore, these lithic assemblages are best
Fig 6. Gruta da Oliveira: Distribution of refit links across layers 15–27 projected on a schematic rendition of the Access Corridor’s stratigraphic sequence. The
vertical black lines denote inter-level refits, the horizontal black lines denote intra-level refits. The red numbers associated with the vertical lines denote the
accumulated number of instances where items from a given layer refit with items in each of the underlying layers. The refits with the Mousterian Cone documenting
long-distance displacement are plotted separately in similar manner. Elevations are in cm below site datum.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 17 / 34
deemed potentially heterogeneous and conservatively removed from further consideration in
terms of human behavior.
Layers 21 and 22 have a strong anthropogenic component, including hearths, which is sug-
gestive of actual human occupation of the Access Corridor—and intra-level refits are indeed
numerous here, which is reflected in these units’ refit success rate (for layer 21, the highest in
the sequence when only the piece-plotted items are considered; Table 1) and in the percentage
Fig 7. Gruta da Oliveira: Horizontal distribution of piece-plotted lithic artefacts from layer 23. The dots represent the (x,y) grid coordinates and they are plotted
on an orthorectified, composite image of the interface between layers 23 and 24 (base of spit A69). Note the concentration in the subsidence-affected area above the
largely openwork e
´boulis filling the connection between the Access Corridor and the Mousterian Cone/Passage of the Column (layers 24–25).
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 18 / 34
of intra plus adjacent refits (86% when layers 21 and 22 are considered together as a single
analytical ensemble; Table 4). Reflecting its stratigraphic interface nature, the upper part of
layer 20 is linked to the stratigraphic units above by a significant number of refits, explaining
why the percentage of intra plus adjacent ones is significantly lower in either the LoP or the
EoP frames of reference (57% and 71%, respectively). When analyzed in detail, however, most
anomalies concern items retrieved in the periphery of the distribution—against the cave walls
and in the openwork e
´boulis at the back end of the trench (Fig 8), where the interface between
layers 19 and 20 was poorly defined and, at the time of excavation, largely established on the
basis of altimetric criteria (i.e., in these peripheral areas the interface plane was defined by pro-
jection from the dip observed in the intact areas, with items falling above being assigned to
layer 19 and items falling below being assigned to layer 20). Excluding from the studied assem-
blages the items retrieved in these problematic parts of the trench therefore ought to suffice to
generate a stratigraphically reliable sample.
Based on the geoarcheological evidence, the artefact component of layers 15–19 was
inferred to result primarily from mass transport over a short distance, the original occupations
having been located a few meters outward [40,42]. The higher proportion of long refits in
these units (especially in layer 16; Table 3) indicates that the validity for human behavior analy-
sis of the archeo-stratigraphic subdivisions made at the time of excavation needs to be critically
Below layer 15 (i.e., for layers 16–19), the refitting patterns are consistent with inward dis-
placement along a slope leading from the entrance to the back of the cavity (Fig 9). They also
show that, if it at all existed, the original micro-stratigraphic organization into different occu-
pation lenses was lost in the transport process. For the purposes of archeological analysis, these
layers’ content should therefore be treated together as a single unit of analysis. Indeed, in the
EoP frame of reference, the proportion of intra plus adjacent refit links obtained for the layers
16–19 ensemble rises to 78%, comparable to that obtained for the better-preserved parts of the
sequence (Table 4). Fig 9 also illustrates well how removing from the plot for columns N and
O the items for which layer 20 is the LoP (to additionally filter out the impact of the distur-
bance displayed by this layer in row 17; Fig 8) makes the separation between the Upper Ensem-
ble (layers 15–19) and the Middle Ensemble (layers 20–22) much clearer: when this is done,
only two refits connect them.
Layer 15, however, stands out within the Upper Ensemble due to the high percentage of
intra plus adjacent refit links—82%, in an EoP frame of reference (Table 4). This is consistent
with the fact that layer 15 represents the last stage of the sedimentary colmatage of the space
left between the large boulders forming the skeleton of the Access Corridor Upper Ensemble;
at that time, the surface of the fill had become largely horizontal, and surface dynamics was
reduced. Coupled with the limited vertical dispersion of the items incorporated in refit sets for
which layer 15 is the LoP (Fig 10), this evidence warrants treatment of the corresponding
stone tool assemblage as an independent unit of analysis. The fact that layer 15’s refit success
rate is comparatively low (Table 1) does not contradict this assessment; it simply reflects the
much larger size of the assemblage (twice the next largest one, that for layer 18), as it is to be
expected that success rates and assemblage size will vary in inverse manner.
Post-depositional dynamics
The vertical projection of refitted items along the axis of deposition illustrates well the mecha-
nisms of post-depositional disturbance at work in Gruta da Oliveira that participated in the
formation of the part of the succession studied here. In Fig 11, we look into the situation in the
Access Corridor and, to highlight the impact of wall effects in column P of the grid (also
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 19 / 34
Fig 8. Gruta da Oliveira: Horizontal distribution of refitted lithic artefacts from layer 20 that reflect downward displacement from overlying units.
The dots represent the (x,y) grid coordinates and they are plotted on an orthorectified, composite image of the base of the first spit (A65) into which layer 20
was subdivided at the time of excavation. Note the concentration in areas susceptible of strong wall effects or consisting of openwork e
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 20 / 34
Fig 9. Gruta da Oliveira: Refit links indicated on the projection of piece-plotted lithic artefacts from the Access Corridor. To
filter out the impact of wall effects, the plot only includes items from grid units N-O/15-17. 1. color codes per stratigraphic unit of
provenience and position of the axes used relative to the excavation grid. 2. projection on the corrY- and z-axes (left) and on the
corrX- and z-axes (right). 3. the same projections with exclusion of the layer 20 material. Note the predominantly sub-horizontal,
stratigraphically consistent orientation of refit links, with very few long-distance, inter-level connections. Removing from the plots
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 21 / 34
apparent in Fig 10), we plotted the stone tools therein separately from those in columns N and
O. The refit lines form a pattern of relatively well-preserved sub-horizontal accumulations in
the latter, except with regards to layers 16–19, in which the observed dip is more important.
This exception agrees with the mode of formation inferred from the geoarcheological data,
which would have favored a telescoping of the deposit inward of the original point of
The horizontal projection of refit links for layers 16–19 and 21–22 (Fig 12) adds support to
the inferences on the dynamics of sediment accumulation based on the geoarcheological data.
For layers 16–19, the preferential direction (NW-SE) of the links is clearly apparent, and many
connections exceed 1 m. For layers 21–22, the orientation of the links is more homogenous
and most connection distances are under 1 m. These observations support that the layers 21–
22 pattern derive from human activity-induced, syn-depositional scatter of debris related to
the occupation of the Access Corridor itself. The preferentially oriented, longer distance con-
nections seen in the case of layers 16–19 are in turn consistent with post-depositional displace-
ment into the Access Corridor of debris related to occupational activities taking place in the
Exterior area of the site.
The horizontal distribution of all the items that go into refits for which layer 15 is the LoP
(column P and rows 17–18 included; Fig 10) shows a concentration in the 27-S Chamber. This
distribution suggests that the latter was the actual emplacement of the human occupation, with
the items retrieved in the Access Corridor primarily representing a peripheral scatter. This is
corroborated by (a) the fact that several refits concern items retrieved quite nearby, over no
more than 1–2 m
of the 27-S Chamber, a case in point being refit R-1133, which conjoins 22
items, 14 of which from grid unit M19 (Fig 5, panel 4), and (b) the absence of refits linking
items retrieved in the Access Corridor alone. The morphology of the cave wall must underpin
this pattern to a significant extent, as, at this elevation, a major salient occupies most of grid
unit O17 and strangulates the communication between those two sections of the cave plan. Fig
10 also highlights the long-distance vertical displacement of a few layer 15 items, while Fig 6
shows how some moved as far down as the Mousterian Cone. Detailed analysis of the projec-
tions against the field documents shows that this displacement takes place in two steps: in the
first, objects syn-depositionally disperse from the 27-S Chamber to the Access Corridor; in the
second, some are post-depositionally displaced downward, along the planes of contact between
the sedimentary deposit and the encasing bedrock found in column P and in row 18.
Diachronic change
To make a preliminary assessment of change in raw-material economy, technology and typol-
ogy across the studied sequence we compared the assemblage from layers 18–19 (combined)
with that from layer 22. These units were chosen to minimize background noise resulting from
the impact of wall effects in the composition of layers situated in intermediate position. Items
<2 cm were excluded. The comparisons concern a total of 798 objects for the layers 18–19
sample and of 564 objects for the layer 22 sample (Figs 1315).
Flint artefacts dominate in the layer 22 sample, followed by quartz and quartzite, while
quartzite dominates in the layers 18–19 sample (Fig 13). Given that the samples are of suffi-
cient size, these changes in raw-material preference must reflect differences in the use of
regional resources related to site function and techno-economic choice.
layer 20, in which row 17 is also affected by post-depositional disturbance, makes the pattern clearer. Elevations are in cm below
site datum.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 22 / 34
Fig 10. Gruta da Oliveira: Refit links of sets for which layer 15 is the LoP (layer of production). The refitted items and associated
refit links are shown on projections of all the piece-plotted finds, including faunal remains, from layer 15. 1. color codes per type of
find and position of the axes used relative to the excavation grid. 2. projection on the horizontal, corrX- vs corrY-axes. 3. projection on
the x- and z-axes, illustrating the concentration of long distance-linked items in column P of the grid. 4. projection on the corrY- and
z-axes, excluding items from column P and row 18 of the grid, illustrating a network of horizontal connections restricted to the 27-S
Chamber. The finds made in layers 26–27 are excluded from the plots. Elevations are in cm below site datum.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 23 / 34
The samples also differ in the type of reduction primarily used. In the layers 18–19 sample,
preferential Levallois debitage is dominant on flint, while quartzite flakes are extracted follow-
ing a centripetal method and only very few are Levallois; the number of Kombewa flakes in
both flint and quartzite is also substantial (Fig 14a and 14b). In the layer 22 sample, flint and
quartzite are reduced in similar manner; the Levallois method is clearly dominant, mostly pref-
erential and, secondarily, recurrent unipolar (Fig 15).
Retouched tools made on flake blanks are rare in the layer 22 sample. Better represented in
the layers 18–19 sample, their overall percentage remains, however, low (8% of flint artefacts).
The blanks selected for retouch are ordinary or naturally-backed flakes, never Levallois ones.
Tool-types are mostly denticulates and notches, followed by flakes with partial retouch, with
some sharpening and re-sharpening being apparent in the layers 18–19 sample (Fig 14c and
14d). The latter sample also contains a limited number of quartzite macro-tools (flake-cleavers
and bifaces; Fig 14f), which are entirely absent from underlying units and, at the site, only
occur in units of the Access Corridor Upper Ensemble (layers 15–19).
In the Portuguese Middle Paleolithic, these Gruta da Oliveira Upper Ensemble levels are
currently the only ones known to contain flake-cleavers and bifaces. One biface has been
reported from the open-air site of Sapateira 2 (Baixo Alentejo), assigned to the Middle Paleo-
lithic, but the archaeological finds made therein are contained in a secondary colluvium [62]
and, therefore, the integrity and dating of the assemblage must be considered an open issue. At
Gruta da Oliveira, refits link all the archeo-stratigraphic units in which these macro-tools are
found, suggesting that the interval represented by their accumulation was rather short. The
absence of such tools from over- and underlying units certainly cannot be due to issues of
assemblage size, and no evidence suggests that it owes anything to functional factors. These
Fig 11. Gruta da Oliveira: Vertical distribution of piece-plotted items from the Access Corridor according to LoP (layer of production). Left. items in column P.
Right. Items in columns N and O, excluding layer 20. Note the numerous long-distance links in column P in contrast with the stratigraphic consistency of the
distributions elsewhere in the trench. Elevations are in cm below site datum.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 24 / 34
Fig 12. Gruta da Oliveira: Horizontal distribution of piece-plotted items from grid units N-O/15-17 and
orientation of the refit links. 1. position of the axes used relative to the excavation grid. 2. projection on the
horizontal, corrX- vs corrY-axes of artefacts included in refit sets whose assigned LoP liesin layers 16 to 19, and radar
graph showing the orientation of the corresponding refit links. 3. projection on the horizontal, corrX- vs corrY-axes of
artefacts included in refit sets whose assigned LoP lies in layers 21–22, and radar graph showing the orientation of the
corresponding refit links.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 25 / 34
patterns plead for the presence of flake-cleavers and bifaces in the Middle Paleolithic of Atlan-
tic Iberia to be a regional industrial marker of wider chrono-stratigraphic significance.
Discussion and conclusions
Our study has shown that, at Gruta da Oliveira, the use for archeological analysis of either the
archeo-stratigraphic units differentiated at the time of excavation or the geo-stratigraphic
Fig 13. Gruta da Oliveira stone tools: Raw-material and technology. 1. representation of the different raw-materials
in the analyzed samples. 2. distribution of technological categories per raw-material (flint and quartzite only) in the
layers 18–19 sample. 3. distribution of technological categories per raw-material (flint and quartzite only) in the layer
22 sample. The categories illustrating the main differences between the two samples are highlighted.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 26 / 34
ensembles defined once excavation was completed and a comprehensive understanding of the
stratification possible needs to be critically assessed. In some cases, the valid unit of archeologi-
cal analysis remains the archeo-stratigraphic unit, in other cases it should be the geo-strati-
graphic ensemble, and it may be necessary in all cases that the assemblages be trimmed into
Fig 14. Gruta da Oliveira stone tools (layers 18–19). a. unretouched flakes (flint and quartzite); b. cores (flint); c. denticulates, and the same with refitted retouch
byproducts (flint); d. denticulates (flint and quartzite); e. refit set R-1011, for which layer 18 is the LoP (layer of production) (quartzite); f. biface and flake-cleaver
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 27 / 34
Fig 15. Gruta da Oliveira stone tools (layer 22). a. unretouched flakes (flint and quartzite); b. refit set R-1000, for which layer 22 is
the LoP (layer of production), illustrating a Kombewa-type reduction (quartzite); c. refit set R-1012, for which layer 22 is the LoP
(left), and a view of the core without the refitted artefacts (right) (quartzite); d. core (flint).
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 28 / 34
samples that exclude items collected in areas where inter-level displacement had a significant
impact. This approach is not novel, and the organization of finds into “Archeological Hori-
zons” whose composition is informed by both stratigraphic provenience and refitting-aided
taphonomic reasoning has been used in the study of both Upper Paleolithic (e.g., Geissenklo¨s-
terle [20] or Le Piage [21]) and Neolithic (e.g., Gruta do Caldeirão [63]) cave and rock-shelter
Given the ubiquitous nature of the period’s reduction methods, applying this approach to
the Middle Paleolithic would seem to be equally, if not more advantageous. Yet, with few
exceptions [32,3536], that is seldom the case, mostly due to the time-consuming nature of
refitting analyses [64], or to the inadequacy of the data. The latter is especially the case when
dealing with old collections, produced by outdated excavation methods, often on-site sorted
(i.e., with discard of the finds devoid of perceived significance), and for which spatial and strat-
igraphic information with the necessary detail may be lacking—e.g., the Passemard excava-
tions at Olha and Isturitz or the Peyrony excavations at Le Moustier and La Ferrassie [6568].
The limited size of the excavation trenches may also prevent the application of a lithic
taphonomy approach. The cave site of Gatzarria (Pyre
´es-atlantiques, France), excavated by
G. Laplace, is a case in point. Here, the Middle Paleolithic deposit is only known after two test
trenches separated by several meters, which hinders the search for refitting links [6971]. It
can also be the case that, when at all re-excavated, old sites are the target of operations limited
to the opening of small trenches or the rejuvenation of extant cross-sections, which poses a
similar problem (see [72] for an exception). Yet, many recent excavations of Middle Paleolithic
stratified sites are amenable to lithic taphonomy analysis, even though the definition of the
analytical units used in the study of the stone tools often remains based solely on the geoarch-
eological unit of provenience.
Thus, our understanding of the Middle Paleolithic is currently affected by a methodological
paradox. On one hand, most sites to which systematic stone tool refitting has been applied to
assess issues of site formation and assemblage integrity are open-air ones. On the other hand,
the criteria upon which the explanation of variation has been framed (e.g. [48]) are derived
from the study of assemblages from cave and rock-shelter sites lacking such a lithic taphonomy
critique. This situation is the more unsatisfactory because, in the Middle Paleolithic, the defini-
tion of industries and technocomplexes is largely based on patterns of association between
traits (technological or typological) that, taken one at a time, are susceptible of being found
across extensive chronological spans. Moreover, post-depositional disturbance is, to a greater
or lesser degree, a universal feature of all stratified sites (the more so in cave and rock-shelter
contexts). Taphonomic analysis of the integrity of the lithics provenanced to the different units
of a given sequence therefore should be a prerequisite for the assessment of their validity as
units of analysis for the archeological study of the human occupations in the context of which
the stone tools were discarded. Coupled with a much-needed improvement in the resolution
of chronometric dating methods, the application of stone tool refitting and associated instru-
ments of lithic taphonomy should constitute the backbone upon which to assess, or rebuild,
the chronostratigraphic schemes upon which the diachronic variability of the European Mid-
dle Paleolithic is currently understood.
In Iberia, the value of this approach has been demonstrated at such sites as the rock-shelters
of El Salt, Quebrada and Romanı
´. At Gruta da Oliveira, our refitting study has shown that the
impact of inter-level displacement was limited and, with the application of the appropriate fil-
ters, can be largely eliminated. The thusly filtered samples are sufficiently large to provide for
reliable analyses of diachronic change, while the better-than-average and indeed quite good
preservation of specific units provides for the use of the entire sample in both technological
and spatial analyses.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 29 / 34
For Gruta da Oliveira specifically, we could conclude the following:
Layer 22, layer 21 and layer 15 are valid units of analysis for the study of short-term activity
in the cave (namely, of intra-site functional variation in relation to associated hearth features
and as revealed by the position of knapping spots or the distribution of use-worn items and
cut-marked or burnt animal bone), while use of layer 20 for the same purpose requires prior
elimination from the sample of items derived from areas in which refits show that wall effects
caused significant contamination by material derived from overlying units.
Within the area of the cave sampled by the excavation trench, the 27-S Chamber was the
main activity area during the accumulation of layer 15.
For the study of long-term change across the sequence, two internally coherent assemblages
can be contrasted, upper (layers 15–19) and lower (layers 20–22)—it being acceptable, given
the refit links and if need be for the convenience of a larger sample size, that layers 23–25 be
added to the layers 20–22 assemblage.
During the time interval of late MIS 5 to early MIS 4 represented by the stratigraphic units
analyzed, raw-material economy and technology shifted from the Levallois reduction of pre-
dominantly flint to the centripetal, mainly non-Levallois reduction of predominantly quartz-
ite and quartz, the latter accompanied by the production and discard of bifaces and flake-
Supporting information
S1 Table. Distribution per archeo-stratigraphic unit of provenience (layer) of individual
items in Gruta da Oliveira (layers 15–27) refit sets. The nature of the refit links, the layer of
provenience (LoP) assignment, and its rationale, are given for each set. The Mousterian Cone
proveniences (layers 26–27) are highlighted. See main text for definitions.
S2 Table. Distribution per analytical unit of provenience (ensemble) of individual items in
Gruta da Oliveira (layers 15–27) refit sets. The nature of the refit links, the ensemble of pro-
venience (EoP) assignment, and its rationale, are given for each set. The Mousterian Cone pro-
veniences (layers 26–27) are highlighted. See main text for definitions.
Marianne Deschamps’s research was supported by a Fyssen Foundation post-doc fellowship.
We acknowledge the help of Henrique Matias with the identification of raw-materials and the
processing of finds and field documents, the pioneer refitting work carried out at Gruta da Oli-
veira by Rui Carvalho, and, last but not the least, the numerous colleagues and volunteers who,
over the years, participated in the Almonda project’s fieldwork. Diego Angelucci read a pre-
liminary version of the manuscript and contributed useful comments. As usual, any errors or
omissions are our own.
Author Contributions
Conceptualization: Marianne Deschamps, João Zilhão.
Investigation: Marianne Deschamps, João Zilhão.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 30 / 34
Methodology: Marianne Deschamps, João Zilhão.
Project administration: João Zilhão.
Writing – original draft: Marianne Deschamps, João Zilhão.
Writing – review & editing: João Zilhão.
1. Spurrell FCJ. On the Discovery of the Place where Palæolithic Implements were made at Crayford.
Quarterly Journal of the Geological Society. 1880; 36(1–4): 544–548.
2. Smith WG. On a Palaeolithic floor at North-east London. Journal of the Anthropological Institute of
Great Britain and Ireland. 1884; 13: 357–386.
3. Leroi-Gourhan A. Le Geste et la Parole. I, Technique et langage. Paris: Michel Albin, coll. "Sciences
d’Aujourd’hui"; 1964.
4. Mauss M. Les techniques et la technologie. Journal de la psychologie. 1948; 41: 250–256.
5. Mauss M. Manuel d’ethnographie. Paris: Payot; 1947.
6. Schlanger N. Le fait technique total, la raison pratique et les raisons de la pratique dans l’oeuvre de Mar-
cel Mauss. Terrain. 1991; 16: 114–130.
7. Leroi-Gourhan A, Bre
´zillon M. L’habitation magdale
´nienne n˚ 1 de Pincevent près Montereau (Seine-et-
Marne). Gallia pre
´histoire. 1966; 9(2): 263–385.
8. Bodu P, Karlin C, Ploux S. Who’s who? The Magdalenian flintknappers of Pincevent (France). In: Czie-
sla E, Eickhoff S., Arts N., Winter D, editors. The Big Puzzle International Symposium on Refitting stone
artefacts. 1990. pp. 143–163.
9. Cziesla E. On refitting of stone artefacts. In: Cziesla E, Eickhoff S, Arts N, Winter D, editor. he Big Puz-
zle International Symposium on Refitting stone artefacts. 1990. pp. 9–44.
10. Karlin C. Analyse d’un processus technique: le de
´bitage laminaire des magdale
´niens de Pincevent
(Seine et Marne). Treballs d’Arqueologia. 1992;(1): 125–161.
11. Almeida F, Brugal J-P, Zilhão J, Plisson H. An Upper Paleolithic Pompeii: Technology, Subsistence and
Paleoethnography at Lapa do Anecrial. From the Mediterranean basin to the Atlantic shore: Papers in
Honor of Anthony Marks. Actas do IV Congresso de Arqueologia Peninsular; Faro: Promontoria Mono-
´fica. 2007; pp. 119–139.
12. Bachellerie F. Quelle unite
´pour le Cha
ˆtelperronien? Apport de l’analyse taphonomique et techno-e
omique des industries lithiques de trois gisements aquitains de plein air: le Baste
´, Bidart (Pyre
Atlantiques) et Canaule II (Dordogne): Ph. D. Thesis, Universite
´de Bordeaux 1. 2011.
13. Cahen D, Moeyersons J. Subsurface movements of stone artifacts and their implications for the prehis-
tory of Central Africa. Nature. 1977; 266: 117–123.
14. Cahen D, Keeley LH, Van Noten FL, Behm JA, Busby CI, Dunnell RC, et al. Stone tools, toolkits, and
human behavior in prehistory [and comments and reply]. Current Anthropology. 1979; 20(4): 661–683.
15. Bordes F. Question de contemporane
´: l’illusion des remontages. Bulletin de la Socie
Franc¸aise. 1980; 77(5): 132–133.
16. Villa P. Conjoinable pieces and site formation processes. American Antiquity. 1982; 47(2): 276–290.
17. Villa P. Torralba and Aridos: elephant exploitation in Middle Pleistocene Spain. Journal of Human Evolu-
tion 1990; 19: 299–309.
18. Villa P, Courtin J. The interpretation of stratified sites: a view from underground. Journal of Archaeolog-
ical Science. 1983; 10: 267–281.
19. Marks AE. Prehistory and Paleoenvironments in The Central Negev, Israel. The Avdat/Aqev Area, Part
3. Dallas: Southern Methodist University; 1983.
20. Hahn J. Die Geißenklo
¨hle im Achtal bei Blaubeuren I. Fundhorizontbildung und Besiedlung im
¨olithikum und im Aurignacien. Stuttgart: Konrad Theiss Verlag; 1988.
21. Bordes J-G. Les interstratifications Cha
ˆtelperronien / Aurignacien du Roc-de-Combe et du Piage (Lot,
France). Analyse taphonomique des industries lithiques; implications arche
´ologiques. Ph. D. Thesis,
´de Bordeaux 1; 2002.
22. Teyssandier N. Les de
´buts de l’Aurignacien en Europe. Discussion àpartir des sites de Geissenklo
terle, Willendorf II, Krems-Hundssteig et Bacho Kiro. Ph. D. Thesis, Universite
´de Paris X-Nanterre;
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 31 / 34
23. Delagnes A. Un mode de production ine
´dit au Pale
´olithique moyen dans l’industrie du niveau 6e du
Pucheuil (Seine-Maritime). Pale
´o. 1993; 5: 111–120.
24. Locht J-L, Swinnen C. Le de
´bitage discoïde du gisement de Beauvais (Oise): aspects de la chaı
ˆne ope
atoire au travers de quelques remontages. Pale
´o. 1994; 6: 89–104.
25. Conard N, Adler D. lithic reduction and hominid Behavior in the Middle Palaeolithic of the Rhineland.
Journal of Anthropological Research. 1997; 53: 147–176.
26. Jo
¨ris O. Der spa
¨olithische Fundplatz Buhlen (Grabungen 1966–69): Stratigraphie, Steinarte-
fakte und Fauna des oberen Fundplatzes. Bonn: Habelt; 2001.
27. Bourguignon L, Sellami F, Deloze V, Sellier-Segard N, Beyries S, Emery-Barbier A. L’habitat mouste
ien de «La Folie» (Poitiers, Vienne): synthèse des premiers re
´sultats. Pale
´o. 2002;(14): 29–48.
28. De Loecker D. Beyond the Site: The Saalian Archaeological Record at Maastricht-Belve
´dère, The Neth-
erlands. Analecta Praehistorica Leidensia. 2004; 35/36:300.
29. Cliquet D, Mercier N, Lautridou J-P, Beugnier V, Bianchini R, Caspar J-P, et al. Un atelier de production
et de consommation d’outils bifaciaux de la fin du Pale
´olithique moyen àSaint-Brice-sous-Ra
(Orne, France) dans son contexte environnemental. Quaternaire. 2009; 20(3): 361–379.
30. Clark A-E. Time and Space in the Middle Paleolithic: Spatial structure and occupation dynamics of
seven open air sites. Evolutionary Anthropology. 2016; 25: 153–163
21486 PMID: 27312187
31. Deschamps M, Clark A-E, Claud E
´, Colonge D, Hernandez M, Normand C. Approche technoe
que et fonctionnelle des occupations de plein-air du Pale
´olithique moyen re
´cent autour de Bayonne
´es-Atlantiques). Bulletin de la Socie
´historique Franc¸aise. 2016; 113(4): 659–689.
32. Vaquero M. The history of stones: behavioural inferences and temporal resolution of an archaeological
assemblage from the Middle Palaeolithic. Journal of Archaeological Science. 2008; 35: 3178–3185.
33. Vaquero M. New perspectives on recycling of lithic resources using refitting and spatial data. Quarta
2011; 58: 113–130.
34. Lo
´pez-Ortega E, Rodrı
´guez XP, Vaquero M. Lithic refitting and movement connections: the NW area of
level TD10-1 at the Gran Dolina site (Sierra de Atapuerca, Burgos, Spain). Journal of Archaeological
Science. 2011; 38(11): 3112–3121.
35. Machado J, Herna
´ndez CM, Mallol C, Galva
´n B. Lithic production, site formation and Middle Palaeolithic
palimpsest analysis: in search of human occupation episodes at Abric del Pastor Stratigraphic Unit IV
(Alicante, Spain). Journal of Archaeological Science. 2013; 40(5): 2254–2273.
36. Machado J, Molina FJ, Herna
´ndez CM, Tarriño A, Galva
´n B. Using lithic assemblage formation to
approach Middle Palaeolithic settlement dynamics: El Salt Stratigraphic Unit X (Alicante, Spain).
Archaeological and Anthropological Sciences. 2017: 9(8): 1–29.
37. Eixea A, Rolda
´n C, Villaverde V, Zilhão J. Middle Palaeolithic flint procurement in Central Mediterranean
Iberia: Implications for human mobility. Journal of Lithic Studies. 2014; 1(1): 103–115 13. Journal of
Lithic Studies. 2014:1(1): 103–115. Available at
38. Romagnoli F, Bargallo
´A, Chaco
´n MG, de Soler BG, Vaquero M. Testing a hypothesis about the impor-
tance of the quality of raw material on technological changes at Abric Romanı
´(Capellades, Spain):
Some considerations using a high-resolution techno-economic perspective. Journal of Lithic Studies.
2016; 3(2). Available at
39. Villaverde V, Eixea A, Zilhão J, Sanchis A, Real C, BergadàM. Diachronic variation in the Middle Paleo-
lithic settlement of Abrigo de la Quebrada (Chelva, Spain). Quaternary International. 2017; 435A: 164–
40. Angelucci D, Zilhão J. Stratigraphy and formation processes of the Upper Pleistocene deposit at gruta
da Oliveira, Almonda karstic system, Torres Novas, Portugal. Geoarchaeology. 2009; 24(3): 277–310.
41. Hoffmann D, Pike A, Wainer K, Zilhão J. New U-series results for the speleogenesis and the Palaeolithic
archaeology of the Almonda karstic system (Torres Novas, Portugal). Quaternary International. 2013;
294: 168–182.
42. Zilhão J, Angelucci D, Aubry T, Badal E, Brugal J-P, Carvalho R, et al. A Gruta da Oliveira (Torres
Novas): uma jazida de referência para o Paleolı´tico Me
´dio da Penı´nsula Ibe
´rica. In: Arnaud J, Martins
A, Neves C, editors. Arqueologia em Portugal 150 anos. Lisboa: Associac¸ão dos Arqueo
´logos Portu-
gueses; 2013. pp. 259–68.
43. Richter D, Angelucci D, Dias M, Prudêncio M, Gouveia M, Cardoso G, et al. Heated flint TL-dating for
Gruta da Oliveira (Portugal): dosimetric challenges and comparison of chronometric data. Journal of
Archaeological Science. 2014; 41: 705–715.
44. Carvalho R, in Zilhão J, Gruta da Oliveira (rede ca
´rsica da nascente do Almonda). Relato
´rio das esca-
vac¸ões de 2006. Unpublished excavation report. 2006.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 32 / 34
45. Laville H, Rigaud J-P, Sackett J. Rock shelters of the Pe
´rigord. New York: Academic Press; 1980.
46. Mellars P. The Neanderthal Legacy. Princeton: Princeton University Press; 1996.
47. Delagnes A, Jaubert J, Meignen L. Les techno-complexes du Pale
´olithique moyen en Europe occiden-
tale dans leur cadre diachronique et ge
´ographique. In: Vandermeersch B, Maureille B, editors. Les
´andertaliens Biologie et culture. Paris: Ed. du CTHS; 2007. pp. 213–229.
48. Jaubert J. Les arche
´quences du Pale
´olithique moyen du Sud-Ouest de la France: Quel bilan un
quart de siècle après Franc¸ois Bordes? 134ème Congrès national des socie
´s historiques et scientifi-
ques, colloque international Franc¸ois Bordes. 29: CTHS; 2013. pp. 235–253.
49. Binford L. Interassemblage variability—the Mousterian and the "functional" argument. In: Renfrew C,
editor. The explanation of culture change: Models in Prehistory. London: Duckworth; 1973. pp. 227–
50. Bordes F. On the chronology and contemporaneity of different Paleolithic cultures in France. In: Ren-
frew C, editor. The Explanation of Culture Change Models in Prehistory. London, Duckworth 1973. pp.
51. Zilhão J, Maurı
´cio J, Souto P. A Arqueologia da Gruta do Almonda. Resultados das escavac¸ões de
1988–89. Actas das IV Jornadas Arqueolo
´gicas (Lisboa, 1990); Lisboa: Associac¸ão dos Arqueo
Portugueses; 1991. pp. 161–171.
52. Zilhão J, Maurı
´cio J, Souto P. Jazidas arqueolo
´gicas do sistema ca
´rsico da nascente do Almonda.
Nova Augusta. 1993; 7: 35–54.
53. Zilhão J, McKinney C. Uranium—Thorium dating of Lower and Middle Paleolithic sites in the Almonda
karstic system (Torres Novas, Portugal). 3ª Reunião do Quaterna
´rio Ibe
´rico Actas; Coimbra: Universi-
dade de Coimbra; 1995. pp. 513–516.
54. Marks A, Monigal K, Zilhão J. The lithic assemblages of the late Mousterian at Gruta da Oliveira,
Almonda, Portugal. In: Zilhão J, Aubry T, Carvalho A F, editors. Les premiers hommes modernes de la
´ninsule Ibe
´rique, Actes du colloque de la commission VIII de l’UISPP, octobre 1998, Vila Nova de
Foz Co
ˆa. Lisboa: Instituto Português de Arqueologia; 2001. pp. 145–154.
55. Zilhão J, Angelucci D, Argant J, Brugal J-P, Carrio
´n J, Carvalho R, et al. Humans and hyenas in the Mid-
dle Palaeolithic of Gruta da Oliveira (Almonda karstic system, Torres Novas, Portugal). In: Baquedano
E, Rosell J, editors. 1a Reunio
´n de cientı
´ficos sobre cubiles de hiena (y otros grandes carnı
´voros) en
los yacimientos arqueolo
´gicos de la Penı
´nsula Ibe
´rica; 21–24 enero 2009; Zona arqueolo
´gica. 2010;7:
56. Badal E, Villaverde V, Zilhão J. Middle Palaeolithic wood charcoal from three sites in south and west
Iberia: biogeographic implications. In: Badal E, Carrio
´n Y, Macı
´as M, Ntinou M, editors. Wood and char-
coal, evidence for human and natural History. Saguntum Extra—13. Valencia: Universitat de Valencia;
2012. pp. 13–24.
57. Matias H. O Aprovisionamento de mate
´rias-primas lı
´ticas na Gruta da Oliveira (Torres Novas). M. Sc.
Thesis, Universidade de Lisboa; 2012.
58. Nabais M. The Neanderthal occupation of Gruta da Oliveira (Almonda karstic system, Torres Novas,
Portugal). Analysis of the burnt bones. In: Maleno P, Lobato C, Lletjo
´s N, A
´lvarez D, Garcı
´a I, Martı
´n F,
et al., editors. Actas de las II Jornadas de Jo
´venes en Investigacio
´n Arqueolo
´gica; 6–8 may 2009;
Madrid: Po
´rtico; 2012. pp. 381–385.
59. Matias H. Raw material sourcing in the Middle Paleolithic site of Gruta da Oliveira (Central Limestone
Massif, Estremadura, Portugal). Journal of Lithic Studies 2016; 3(2). Available at
60. Angelucci D. The Geoarcheological Context. In: Zilhão J, Trinkaus E, editors. Portrait of the Artist as a
Child. The Gravettian Human Skeleton from the Abrigo do Lagar Velho and its Archeological Context.
22. Lisboa: Instituto Português de Arqueologia2002. pp. 58–91.
61. Carvalho R, in Zilhão J, Gruta da Oliveira (rede ca
´rsica da nascente do Almonda). Relato
´rio das esca-
vac¸ões de 2010. Unpublished excavation report. 2010.
62. Cura S. Matière première et variabilite
´technologique au Pale
´olithique moyen portugais: l’exemple du
site Sapateiros 2 (Baixo Alentejo, Portugal). M. A. Thesis, University Paris 1 Panthe
63. Zilhão J. Gruta do Caldeirão. O Neolı
´tico Antigo. Lisboa: Instituto Português do Patrimo
´nio Arquitecto
nico e Arqueolo
´gico; 1992.
64. Laughlin J, Kelly R. Experimental analysis of the practical limits of lithic refitting. Journal of Archaeolog-
ical Science. 2010; 37: 427–433.
65. Passemard E. Le Mouste
´rien àl’abri Olha en Pays-Basque. Revue Lorraine d’Anthropologie.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 33 / 34
66. Passemard E. La caverne d’Isturitz en Pays Basque. Pre
´histoire. Paris: Presses Universitaires de
France. 1944.
67. Peyrony D. Le Moustier: ses gisements, ses industries, ses couches arche
´ologiques. Revue anthropo-
logique. 1930; 40: 48–76, 155–176.
68. Peyrony D. La Ferrassie: Mouste
´rigordien-Aurignacien. Pre
´histoire. 1934; 3:1–92.
69. Laplace G, Saenz de Buruaga A. Typologie analytique et structurale des complexes du Mouste
´rien de
la grotte de Gatzarria (Ossas-Suhare, Pays Basque) et de leurs relations avec ceux de l’abri Olha 2
(Cambo, Pays Basque). Pyrenae. 2002–2003; 33–34: 81–163.
70. Deschamps M. La diversite
´culturelle au Pale
´olithique moyen re
´cent: le Vasconien et sa signification au
sein des faciès mouste
´riens. Ph. D. Thesis, Universite
´de Toulouse 2; 2014.
71. Deschamps M. Late Middle Palaeolithic assemblages with flake cleavers in the western Pyrenees: The
Vasconian reconsidered. Quaternary International. 2017; 433B: 33–49.
72. Thomas M. Etude techno-e
´conomique et taphonomique de palimpsestes d’occupations: l’exemple de
la couche H de l’abri du Moustier (Dordogne, France). M. A. Thesis, Universite
´de Toulouse 2; 2017.
Gruta da Oliveira stone tool refitting
PLOS ONE | February 16, 2018 34 / 34

Supplementary resources (2)

... On this matter, many studies have addressed these questions around the world in the last 40 years and from different perspectives (e.g. Butzer 1982;Villa 1982;Hofman 1986;Schick 1986;Schiffer 1987;O'Connell 1987; Barton and Clark 1993;Morrow 1996;Dibble et al. 1997;Zilhão and d'Errico 1999;Bordes 2002;Lenoble and Bertran 2004;Morin et al. 2005;Bailey 2007;Texier 2009;Goldberg et al. 2009;Staurset and Coulson 2014;Aubry et al. 2014;Marwick et al. 2017;Deschamps and Zilhão 2018;Bergadà et al. 2018;Sanchis et al. 2019), although geoarchaeology is the discipline that has analysed them more widely and continuously. ...
... This type of material is ubiquitous and abundant in many deposits, especially those with Palaeolithic occupations, justifying the high suitability of this approach for being applied in most sites. The main tools of lithic taphonomy are refitting (Villa 1982;Bordes 2000;Staurset and Coulson 2014;Deschamps and Zilhão 2018), second-order refits (Petraglia 1992;Bordes 2002;Pollarolo et al. 2010), techno-typological characterisation (Dibble 1995;Tsanova 2008;Bachellerie 2011), analysis of surface alterations (Schoville 2014;Galland et al. 2019), granulometry (Bertran et al. 2006b;de la Torre et al. 2018) and fabric analysis (Lenoble and Bertran 2004;Benito-Calvo et al. 2009). The combined application of some of them allows constructing a more reliable taphonomic reading of the archaeological record. ...
... Anyway, the minor character of long vertical connections in the analysed sequence points to the absence of generalised alterations that displace remains vertically over long distances. Thus, in Cendres, this type of movement seems more isolated, differently from what was detected in the karstic deposit of Gruta da Oliveira (Deschamps and Zilhão 2018) or in sand deposits such as Gombe Point (Cahen 1976), Terra Amata (Villa 1982) or White Paintings Shelter (Staurset and Coulson 2014). In relation to the latter three examples, maybe the type of sediment of Cendres deposit, with a prominent component of silts and clays (Villaverde et al. 2019, p. 94), added to the remarkable humidity of the matrix, provides higher compaction of the matrix and may have stemmed post-depositional movements of the remains. ...
Full-text available
The formation processes of archaeological deposits determine the potential information that could be obtained in a site, so they influence the interpretations about past societies. This paper presents the lithic taphonomy study of the Aurignacian, Gravettian and Solutrean levels of Cova de les Cendres. For this purpose, lithic refits and typological characteristics of the lithic industry are analysed in statistical and spatial terms. The use of density maps of refitting connection lines is an innovative procedure that stands out in this study. Most of the connections are intra-level and long vertical connections are very scarce. In addition, lithic morphotypes characteristic of each period show a coherent archaeostratigraphical distribution. Results suggest that most of the levels preserve some degree of integrity. Furthermore, the post-depositional alterations had a deeper impact horizontally, whereas their influence in the vertical sense is not that deep. The spatial analysis shows vertical and horizontal variations throughout the sequence, highlighting spatial information as an essential aspect to reach a complete image of the deposits in these critical analyses. This work underscores the necessity to extend these taphonomic studies to other Upper Palaeolithic sequences and cave contexts to make more accurate interpretations about cultural evolution.
... Despite the absence of flint, technological data present some similarities with Gruta da Oliveira's layers 15 to 25, the reference for Middle Palaeolithic occupation of Central Portugal that has been recently redated by U-series using speleothems, coupled with new luminescence ages for sediment infill. Data show that the site's ~6-m-thick archaeological stratigraphy dates entirely within a <30 ka interval, spanning substages 5a-5b of MIS 5, confirming a shift from the centripetal Levallois reduction to discoid and Kombewa debitage during the time interval of late MIS 5 to early MIS 4 (Deschamps and Zilhão, 2018) and that flake-cleavers and bifaces defining the Vasconian facies are restricted to a short interval correlated with Greenland Stadial (GS) 22, 85.1-87.6 ka ago (Zilhão et al., 2021a). ...
... Discoid reduction methods deduced from the lithic assemblage found throughout the GFU 5 Middle Palaeolithic layers in the 3 areas excavated at Cardina-Salto do Boi are widespread in Iberian Late Middle Palaeolithic assemblages and dated to MIS 5a and MIS 4 (Deschamps and Zilhão, 2018). Along the GFU 5 the selection of local translucent quartz, the reduction sequence used, and morphology of the flakes obtained are similar in the three areas where GFU 5 has been excavated (Fig. 13). ...
Hunter-gatherer societies mobility has been interpreted as a dialectical interplay of social and environmental factors. Demography and social network restriction have been pointed out as potential factors for the demise of Neanderthal and to differ them from anatomically modern human. To reconstruct land use and social network we investigate Middle and Upper Palaeolithic lithic remains from an open-air site located in Iberian hinterland, spanning from MIS 5 to 3. In a geological environment with a variety of quartz veins but no available flint or silcrete sources, data reveal common patterns through Neanderthal occupations, and substantial differences interpreted as distinct subsistence strategies and social networks of the two populations.
... Graph based on Monte Carlo methods showing the presence of artefacts (d), clasts (e) and artefacts-clasts (f) and 999 random simulations (grey) according to slope at OC1 level. Schick, 1986, Pope, 2002, Deschamps and Zilhao, 2018, López-Ortega, et al., 2019. At the levels studied, we have mainly observed examples of conjoins from pieces that have broken through postdepositional processes, although there are also some examples that could be linked to on-site knapping activities. ...
The analysis of formation processes (e.g., degree of surface alterations, size ranges, orientation patterns, fabric features, density, spatial patterns, etc.) in archaeological sites provide relevant information about the type and intensity of sedimentary disturbances. This type of analysis is particularly important in Palaeolithic sites, especially when the stratigraphic context suggests certain sedimentary disturbances. This study follows a multivariate approach to the Acheulean site (pre-dating Marine Isotopic Stage -MIS- 5) at Arbo (Lower Miño basin, Spain). The two layers analyzed (levels OC1 and OC2) with archaeological remains (exclusively lithic artefacts) show matrix-supported gravel and cobble facies, which are compatible with extensive sedimentary disturbances. The results obtained demonstrate that sedimentary processes altered the archaeological assemblages in different ways, with some sorting and surface abrasion, in addition to a significant alteration of initial spatial patterns. However, the degree of disturbances observed is not sufficient to invalidate the documented techno-typological information, which is representative of the patterns of human behaviour at the site. This reinforces the importance of this type of analysis as a necessary step in the interpretation of any archaeological site, especially in Palaeolithic ages.
... The Middle and Upper Palaeolithic of Gruta do Caldeirão manifestation of the so-called "wall effect" [72]. Such effects often entail the downward migration of small items, and it can be hardly coincidental that Neolithic intrusions into apparently intact areas of layers Fa and Fb were only found along the north wall of the cave in O12 and P12 (i.e., adjacent to the Fb1 feature). ...
Full-text available
Gruta do Caldeirão features a c. 6 m-thick archaeological stratification capped by Holocene layers ABC-D and Ea, which overlie layer Eb, a deposit of Magdalenian age that underwent significant disturbance, intrusion, and component mixing caused by funerary use of the cave during the Early Neolithic. Here, we provide an updated overview of the stratigraphy and archaeological content of the underlying Pleistocene succession, whose chronology we refine using radiocarbon and single-grain optically stimulated luminescence dating. We find a high degree of stratigraphic integrity. Dating anomalies exist in association with the succession’s two major discontinuities: between layer Eb and Upper Solutrean layer Fa, and between Early Upper Palaeolithic layer K and Middle Palaeolithic layer L. Mostly, the anomalies consist of older-than-expected radiocarbon ages and can be explained by bioturbation and palimpsest-forming sedimentation hiatuses. Combined with palaeoenvironmental inferences derived from magnetic susceptibility analyses, the dating shows that sedimentation rates varied in tandem with the oscillations in global climate revealed by the Greenland oxygen isotope record. A steep increase in sedimentation rate is observed through the Last Glacial Maximum, resulting in a c. 1.5 m-thick accumulation containing conspicuous remains of occupation by people of the Solutrean technocomplex, whose traditional subdivision is corroborated: the index fossils appear in the expected stratigraphic order; the diagnostics of the Protosolutrean and the Lower Solutrean predate 24,000 years ago; and the constraints on the Upper Solutrean place it after Greenland Interstadial 2.2. (23,220–23,340 years ago). Human usage of the site during the Early Upper and the Middle Palaeolithic is episodic and low-intensity: stone tools are few, and the faunal remains relate to carnivore activity. The Middle Palaeolithic is found to persist beyond 39,000 years ago, at least three millennia longer than in the Franco-Cantabrian region. This conclusion is upheld by Bayesian modelling and stands even if the radiocarbon ages for the Middle Palaeolithic levels are removed from consideration (on account of observed inversions and the method’s potential for underestimation when used close to its limit of applicability). A number of localities in Spain and Portugal reveal a similar persistence pattern. The key evidence comes from high-resolution fluviatile contexts spared by the site formation issues that our study of Caldeirão brings to light—palimpsest formation, post-depositional disturbance, and erosion. These processes. are ubiquitous in the cave and rock-shelter sites of Iberia, reflecting the impact on karst archives of the variation in climate and environments that occurred through the Upper Pleistocene, and especially at two key points in time: between 37,000 and 42,000 years ago, and after the Last Glacial Maximum. Such empirical difficulties go a long way towards explaining the controversies surrounding the associated cultural transitions: from the Middle to the Upper Palaeolithic, and from the Solutrean to the Magdalenian. Alongside potential dating error caused by incomplete decontamination, proper consideration of sample association issues is required if we are ever to fully understand what happened with the human settlement of Iberia during these critical intervals, and especially so with regards to the fate of Iberia’s last Neandertal populations.
... Due to the importance of post-depositional processes in archaeological site formation, most spatial studies focus on the influence of post-depositional processes on the dispersion of archaeological artefacts (Benito-Calvo & de la Torre, 2011;Boschian & Saccà, 2010;Sánchez-Romero et al., 2016). In addition, more attention is paid to vertical displacements for the purpose of interpreting stratigraphic dynamics or temporal events, while horizontal intra-site spatial interpretations are often neglected (Deschamps & Zilhão, 2018). ...
Conference Paper
Humans evolved specializations to integrate tools into their cultural, perceptive and cognitive systems. Tools extend cognitive functions beyond the brain and directly improve the range of our cognitive skills. A characteristic of human beings is their hand-eye coordination, associated with a specialized visuospatial system. In fact, parietal lobes, larger and more complex in modern humans, are involved in the management of the relationships between brain, body and environment. Tools activate brain motor regions even during passive viewing. When a tool is touched, it is integrated into the body scheme, expanding the peripersonal space. Recently, different disciplines have adopted an embodied cognition perspective arguing that physical tools should be considered as a functional part of the cognitive network. Vision is the dominant source of sensory information in primates, channelling action and body-environment relationships. We applied eye-tracking technology to investigate visual perception during human interaction with Lower Paleolithic stone tools. We quantified visual attention during a free exploration of stone tools in peripersonal space and during tool physical manipulation. Our results suggest that choppers are usually more explored at the upper region and cortex, while handaxes receive more attention at the base and edges. We can consider whether choppers, displaying a simpler morphology, may require less attention associated with the grasping strategy, while handaxes, with a more complex profile and more grasping and use possibilities, may need a finer exploration of its base. That differences during visual exploration are the same whether or not the tools are physically manipulated, suggesting haptic exploration simply serves visual perception, without influencing the visual exploration scheme. The analysis of visual behaviour associated with these affordances can supply information on the early steps of this brain-body-tool interaction, evidencing common factors as well as type-specific perceptual differences
... Refitting is a very effective approach to identify the intra-site activities and the social organizations of the pre-historic human groups. The importance of refitting analysis has already been tested, in terms of archaeostratigraphy and the site formation studies, as it provides crucial information about the integrity of an assemblage, the types and degree of post-depositional processes affecting it, and in the end, its temporal resolution (Ashton 2004;Bargalló et al. 2016;Deschamps and Zilhão 2018;Hofman 1986;Villa 1982). Undoubtedly, in terms of the social organization of the Paleolithic human groups, spatial interrogation with refitting method has been applied to several sites, being the most famous some Magdalenian settlements as the well-known sites in the Paris basin such as Verberie (Audouze 1988;Audouze and Enloe 1997;Audouze et al. 1981), Pincevent (Karlin and Julien 2019;Leroi-Gourhan and Brézillon 1966), and Étiolles (Caron-Laviolette et al. 2018;Olive 1988;Olive et al. 2019;Pigeot 1987). ...
Full-text available
In this manuscript, we present the first systematic refitting results of the small-scale Middle Pleistocene (MIS11) rock shelter site of La Cansaladeta. The lithic materials that have been recovered from the archaeological levels E and J were the main study materials. These levels were investigated regarding spatial pattern analysis and analyzed with auxiliary methods such as quantitative density mapping demonstration and technological analysis of the lithic clusters. Thus, the spatial patterns of the two levels were compared and discussed, in terms of connections, clusters, and movement of the lithic elements. Undoubtedly, the well preservation of the archaeological levels offered a great opportunity for the interpretation of the spatial patterns in a high-resolution perspective. La Cansaladeta has not been paid attention adequately so far may be due to the small dimension of the excavation surface or to the scarcity of faunal record. Our results show that small-scale sites without long-distance refit/conjoin connections can provide significant spatial information. Indeed, if the sites have very well-preserved archaeological levels, the absence of long connections can be supported by the auxiliary methods.
... The karst system is situated approximately 40 km inland between 75 and 150 m.a.s.l., oriented in a NE-SW direction and is home to a network of cave passages and archaeological sites, namely Galeria da Cisterna, Gruta da Oliveira, Gruta da Aroeira and Galeria das Lâminas (GdL) (Figure 1). (Daura et al., 2017;Deschamps and Zilhão, 2018;Marks et al., 2001;Trinkaus et al., 2007Trinkaus et al., , 2011Zilhão, 2009;Zilhão et al., 1991Zilhão et al., , 2010 (Badal et al., 2012). ...
Full-text available
The European climate during the Holocene period is characterised by frequent changes of temperature and precipitation. The North Atlantic plays a major role as a driver for European climate and is a dominant precipitation source, particularly for the western European and north African realm. Atmospheric pressure gradients over the Atlantic (North Atlantic Oscillation, NAO), Atlantic circulation patterns (Atlantic Multidecadal Oscillation, AMO) or positioning of the Atlantic jet stream have been suggested to be responsible for precipitation patterns across western Europe. However, proxy data provide an inconsistent picture on how precipitation responds to changes in the Atlantic realm such as changes of Atlantic temperature (IRD), atmospheric pressure (NAO), water circulation (AMO) or the jet stream. Here we present a record of speleothem-based winter precipitation amount from Portugal. The record covers most of the Holocene and demonstrates that wetter conditions were synchronous in western and southern Iberia during early and mid Holocene. The record also shows a correlation between increased winter precipitation amount in western Iberia and Atlantic cooling, evidenced by Bond events, between 10 and 4 ka.
... Due to the importance of post-depositional processes in archaeological site formation, most spatial studies focus on the influence of post-depositional processes on the dispersion of archaeological artefacts (Benito-Calvo & de la Torre, 2011;Boschian & Saccà, 2010;Sánchez-Romero et al., 2016). In addition, more attention is paid to vertical displacements for the purpose of interpreting stratigraphic dynamics or temporal events, while horizontal intra-site spatial interpretations are often neglected (Deschamps & Zilhão, 2018). ...
In situ archaeological assemblages are scarce. Most artefacts are affected by post-depositional processes, which complicate the identification of human factors involved in the formation of lithic workshops, such as hand laterality, knapping positions, and knapping expertise. In this regard, experimental archaeology can reproduce past processes to test present hypotheses. In this study, experimental knapping processes were carried out to examine how techniques and methods learned by experience can affect waste flake assemblages. Our results show that, in addition to knapping position, experience affects both flake measurements and spatial distribution. The techniques applied, the knappers’ skill and the control of the application of force all affect spatial flake distribution. However, flakes do not exhibit any preferential orientation, contrary to those affected by some post-depositional factors. The experimental design and results could contribute to the ability to discern Palaeolithic in situ assemblages as well as some human factors that affect their formations.
Based on previous radiocarbon and U-series (Diffusion/Adsorption) dating of bone samples, the Middle Palaeolithic has been thought to persist at Gruta da Oliveira until ∼37 thousand years (ka) ago. New U-series ages for stratigraphically constraining speleothems, coupled with new luminescence ages for sediment infill, show that the site’s ∼6 m-thick archaeological stratigraphy dates entirely within a <30 ka interval spanning substages 5a-5b of Marine Isotope Stage (MIS) 5. Significant technological change is observed across the sequence, akin to that seen in the Upper Palaeolithic over similar timescales. Flake-cleavers and bifaces, normatively definitional of the Vasconian facies, are restricted to a short interval correlated with Greenland Stadial (GS) 22, 85.1–87.6 ka ago. In cave and rock-shelter sites of southern and western Iberia, intact archaeological deposits securely dated to the ∼37–42 ka interval remain elusive. Geological dynamics (e.g., erosion, sedimentation hiatuses, palimpsest formation) and human adaptive responses to climate-driven environmental change (e.g., abandonment of now forest-covered low- and mid-altitude karst areas, concentration of settlement in alluvial plains and coastal settings) are possible explanations for this pattern.
This chapter reviews the production and interpretation of chipped- and ground-stone tools, including the chaînes opératoires that describe processes from the acquisition of raw materials through the manufacture and recycling of these types of tools. It describes the conventional terms for the “anatomy” of lithic flakes and some of the main types of ground-stone tools, and outlines some of the attributes and methods that archaeologists commonly study in their attempts to infer lithic technology and tool use. There are brief reviews of use-wear and non-use alteration of tools, of style in tools, and ways to ensure the validity and reliability of stone-tool analyses.
Full-text available
Technological changes have been identified in several European Middle Palaeolithic sites. Specifically, the turnover in discoid and Levallois knapping methods has traditionally been explained by raw material constraints that are usually related to foraging areas and mobility strategies of Neanderthal groups. While Levallois production requires high homogeneous blocks, predominant discoid techno-complexes have generally been interpreted as better adapted to the scarcity of high quality raw material, not only for the lowest degree of control in products morphology, but also for their multitask characteristics. Nevertheless, the impact of the quality of raw material has never been systematically studied. Furthermore, technological analyses usually consider the lithic assemblage as a whole and do not dissect assemblages to identify single events, which are units that are needed to interpret relationships between technological organisation, human mobility, economic strategies, and settlement patterns. Here, we present an application of technological analysis with a high-resolution approach to investigate, in detail, how raw material quality affected production and how Neanderthals managed the low quality of Sant Martí de Tous chert within Levallois and discoid concepts. We used Raw Material Units and refits as units of analysis with a diacritical approach. The results suggested that the Levallois organisation of the reduction sequence in layer O included a phase of selection of the block and its systematic cutting-down, as well as quite standardised productive procedures and a high fragmentation of the productive sequence within the landscape. In layer M discoid sequences showed a high internal variability as a response to raw material constraints, and most of the production was usually manufactured at the site. Data implied that factors others than the quality of raw material determined the technological turnover at Abric Romaní, suggesting that social organisation and settlement patterns have most probably played a more significant role than foraging area.
Full-text available
A few meters above the karstic spring of the Almonda River, there is a cave entrance that gives access to an extensive network, several kilometers long, of underground galleries. In the first few meters of that cave, shallow archeological deposits from different time periods (Upper Paleolithic and Neolithic; Copper, Bronze and Iron Ages) have been known to exist since the 1930's. As a result of a program of systematic speleoarcheological exploration of the system, began in 1989 and continuing, Lower and Middle Paleolithic deposits were also discovered (ZILHAO et al. 1991, 1993). These earlier deposits are located in two different areas pig. 1): the fossil galleries above the spring; and the EVS zone, several hundred meters upriver, in deep galleries subject to Winter flooding. Uranium-Thorium dates for some of these contexts have been obtained (Table 1). In this paper they will be presented and briefly discussed.
Full-text available
The cave site of Gruta da Oliveira is located in the Almonda karst system, at the interface between the Central Limestone Massif of Portuguese Estremadura (CLM) and the adjacent Sedimentary Basin of the River Tagus (TSB). The cave presents a stratification dated to ~37-107 ka containing hearth features, Neanderthal skeletal remains, as well as fauna, microfauna and wood charcoal remains. The lithic assemblages are large and feature a diverse range of raw materials. Knappable lithic raw materials in primary, sub-primary and secondary position in the CLM and the TSB were systematically surveyed and sampled. The characterization of the geological samples was carried out at both the macro- and the microscopic scales and data were systematized under the petroarcheological and “evolutionary chain of silica” approaches. The study of the lithic assemblage from layer 14 (dated to the ~61-93 ka 95.4% probability interval by TL) indicates that the Gruta da Oliveira Neanderthals used quartzite, quartz and flint from sources located less than 30 km away in both the CLM and the TSB.
Full-text available
Full-text available
A Gruta do Almonda é uma estação arqueológica conhecida desde há cerca de 50 anos, data em que foi posta a descoberto a entrada pela qual se faz o acesso ao seu interior. Trata-se de uma surgência fóssil do Rio Almonda (embora em Invernos excepcionalmente pluviosos tenha já funcionado igualmente como saída de águas), cuja nascente actual se encontra cerca de cinco metros mais abaixo e é represada pela Fábrica de Papel da Renova, a cuja laboração fornece a água necessária. Pela referida entrada acede-se a uma extensa rede de galerias subterrâneas, já reconhecidas numa extensão de cerca de 5 km (fig. I) , mas era somente nas primeiras dezenas de metros que, até a realização das campanhas de escavação realizadas em 1988 e 1989, se conhecia a existência de vestígios arqueológicos (Paço et a/. 1947; Guilaine e Ferreira 1970). Os trabalhos realizados nos Últimos dois anos tiveram a sua origem numa descoberta realizada por elementos da Sociedade Torrejana de Espeleologia e Arqueologia: o achado nesta porção inicial da galeria de entrada de algumas peças solutrenses cujo contexto se tornava imprescindível estabelecer (Maurício 1988; Zilhão 1988). A realização dos trabalhos veio porém dar igualmente origem a aquisição de novos dados referentes as ocupações já documentadas pelas escavações de finais dos anos 30 (Neolítico antigo, Idade do Bronze e Idade do Ferro), e a descoberta de ocupações do Paleolítico Inferior e Médio em zonas da gruta cuja importância arqueológica era até então desconhecida. São estes resultados que ora se apresentam de forma resumida e preliminar. 1. Galeria da Nascente (zonas AMD1, AMD2 e AMD3) Na parte inicial da galeria realizaram-se trabalhos em quatro áreas diferentes: a zona do primeiro alargamento da galeria, situada a cerca de 10m da entrada (AMD1, quadrados L-M/10-14); a zona adjacente a pequena sala que foi escavada por Afonso do Paço, e onde haviam sido descobertas as peças solutrenses (AMDI, quadrados M115-20); a zona imediatamente a seguir ao ressalto rochoso que separa esta parte inicial da galeria da conduta que dá acesso à cisterna e a continuação da rede cársica (AMD2, quadrados F-Gl18-23); e, finalmente, a zona AMD3, correspondente a última bolsa preenchida com sedimentos onde eram visíveis materiais arqueológicos a superfície (fig. 2). O carácter intacto dos depósitos escavados nestas zonas sugere que as escavações dirigidas por Afonso do Paço se terão efectivamente restringido a área indicada na fig. 1, embora seja provável que o conjunto dos materiais provenientes desses trabalhos inclua também peças recolhidas a superfície em toda a extensão da galeria. A evacuação dos sedimentos escavados em 1988-89 foi possibilitada pela instalação de um sistema de carris para circulação de uma pequena vagonete, sem o que o transporte dos baldes para o exterior da gruta seria extremamente difícil, dadas as reduzidas dimensões da galeria nos seus 6-7m iniciais. A crivagem dos baldes da zona AMD1 foi realizada a seco, junto a entrada da gruta, com crivos de malha de 2mm. Os baldes da zona AMD2, dada a sua enorme riqueza em objectos de adorno de pequenas dimensões, foram integralmente crivados a água na barragem da Renova, para onde eram enviados através de um sistema de cabo e roldana, sendo igualmente de 2mm a malha dos crivos usados para o efeito. Na zona AMD1, os trabalhos revelaram uma estratigrafia de reduzida potência (cerca de 90cm) mas em que era possível distinguir trêsgrandes unidades. A primeira (camada I) , com uma espessura média de cerca de 40cm, correspondia a um preenchimento holocénico contendo materiais arqueológicos relacionáveis com as Idades do Cobre e do Bronze (cerâmica
Full-text available
The spatial structure of archeological sites can help reconstruct the settlement dynamics of hunter-gatherers by providing information on the number and length of occupations. This study seeks to access this information through a comparison of seven sites. These sites are open-air and were all excavated over large spatial areas, up to 2,000 m2, and are therefore ideal for spatial analysis, which was done using two complementary methods, lithic refitting and density zones. Both methods were assessed statistically using confidence intervals. The statistically significant results from each site were then compiled to evaluate trends that occur across the seven sites. These results were used to assess the “spatial consistency” of each assemblage and, through that, the number and duration of occupations. This study demonstrates that spatial analysis can be a powerful tool in research on occupation dynamics and can help disentangle the many occupations that often make up an archeological assemblage.
Full-text available
The excavation of the Magdalenian site of Vale da Mata also yielded an ensemble of Neolithic and Copper Age finds. The spatial distribution of the Neolithic material shows that it formed a separate, discrete cluster, and its composition suggests an occupation of residential nature. The small set of Copper Age finds comprises items lacking secure or specific provenience information; they may have been surface-collected in the surrounding area. In good agreement with the evidence for deforestation and concomitant emergence of agricultural fields seen in the regional paleoenvironmental archives after 5400 cal BC, the existence of an Early Neolithic settlement of the lower Sizandro valley is confirmed.