Content uploaded by Antonieta Jerardino
Author content
All content in this area was uploaded by Antonieta Jerardino on May 01, 2016
Content may be subject to copyright.
Shell density as proxy for reconstructing prehistoric aquatic resource
exploitation, perspectives from southern Africa
Antonieta Jerardino
Research Professor, ICREA/Department of Experimental & Health Sciences, Universitat Pompeu Fabra, CaSEs Research Group, Ramon Trias Fargas 25–27, 08005 Barcelona, Spain
Research Associate, Department of Anthropology & Archaeology, University of South Africa(UNISA), PO Box 392, Pretoria, South Africa
abstractarticle info
Article history:
Received 18 April 2015
Received in revised form 1 June 2015
Accepted 5 June 2015
Available online 13 June 2015
Keywords:
Shell midden
Shellfish collection
Site formation
Settlement pattern
Deposition rates
Shell taphonomy
Prey transport
Aquatic resource use has gained enormous attention in recent years, particularly in terms of its role in human
evolution. Significant strides are being made regarding the potential nutritional significance of aquatic foods
for hominin diets, and explicit conceptual frameworks for understanding the evolutionary context of coastal
adaptations are also being developed. Finding out when and where systematic use of aquatic resources took
place and what constitutes a well-developed coastal adaptation requires a corpus of data that can offer insights
into theorganization of marine resource procurement. Theearliest evidencefor the exploitation of marine coastal
habitats in the world is found in southern-most Africa (≤164 ka), where the mostevident and abundant material
expression of such foraging adaptation is revealed by marine mollusc shells found in mid to late Pleistocene
archaeological sites. Hence, it becomes imperative to understand what can be meaningfully inferred from the
variable quantities of such a significant component of early marine aquatic exploitation. This paper approaches
the issue of aquatic resource use by investigating one of the most frequently employed quantitative measure
for inferring such behaviour, namely shell density. Holocene assemblages from the South African west coast
are used as a case study. It is found that shell densitiescan be misleading for inferring intensity of coastalresource
use in the absence of palaeoshorelinereconstruction and when deposition rates are assumed to remain constant
and when geomorphological and taphonomic contexts are unknown.Overall, deposition rates of marine shells is
a better proxy for comparison between sites and through time, and thus for gaining insight into past marine
coastal adaptations.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
Aquatic resources have been viewed as marginal until not too long
ago, and the widespread appearance of shell middens along shorelines
postdating the Last Glacial Maximum (LGM) was taken as indicative of
their late systematic use once human population levels had increased
substantially and apparently more productive terrestrial foods were
depleted (i.e., Osborn, 1977). But archaeologists know by now that
coastal resources have been exploited regularly since at least the Middle
and Late Pleistocene as evidenced by the presence of shell-bearing sites
in southern Africa (164 ka; Avery et al., 2008; Jerardino and Marean,
2010; Langejans et al., 2012; Henshilwood et al., 2014; Kyriacou et al.,
2015) and Europe (150 ka; Colonese et al., 2011; Cortés-Sánchez et al.,
2011). Hence, one of the main issues now revolves around the when
and where systematic exploitation of coastal resources took place. An-
swering these questions requires reflecting upon definitions of what
systematic exploitation of coastal resources is and how to go about de-
tecting such an adaptation in the archaeological record. Methodological
approaches for inter-assemblage and site comparisons on quantitative
grounds thus need to be developed. An additional and important chal-
lenge to this endeavour in southern Africa, as elsewhere, is that much
of this record has been lost or is unavailable due to successive postgla-
cial sea level rise and flooding of the world's continental shelves
(Bailey and Flemming, 2008; Fisher et al., 2010). When this record has
survived thanks to steep bathymetry or other fortunate factors, its ma-
terial expression could be nonetheless seriously compromised or ob-
scured due to forager transport-mediated decisions. An increasing
distance between campsites and shoreline is likely to have affected
the choices of what and how much to bring back to campsites asethno-
graphic and archaeological studies show (e.g., Thackeray, 1988; Bird
and Bliege Bird, 1997; Marean and Cleghorn, 2003; Lupo, 2007;
Thomas, 2007; Dusseldorp and Langejans, 2013). This is an issue that
cannot be dismissed when studying past shellfish collecting strategies
in the context of changing coastline configurations as a result of sea
level change.
Systematic exploitation of coastal resources has been equated with
relatively large, localised and dense shell middens in association with
faunal remains, artefacts and sometimes site features (Parkington,
2003; Jerardino, 2010a; Will et al., 2015). But it is not immediately
apparent when shell dumps are dense and large enough, and/or suffi-
ciently localised to reflect systematic exploitation of coastal resources.
Journal of Archaeological Science: Reports 6 (2016) 637–644
E-mail addresses: antonieta.jerardino@upf.edu,amsjerardino@gmail.com.
http://dx.doi.org/10.1016/j.jasrep.2015.06.005
2352-409X/© 2015 Elsevier Ltd. All rights reserved.
Contents lists available at ScienceDirect
Journal of Archaeological Science: Reports
journal homepage: www.elsevier.com/locate/jasrep
Alternative views propose that the sporadic use of coastal resources is
not an indication of a coastal adaptation and that rather human life-
ways ought to be transformed significantly (i.e., subsistence and settle-
ment patterns, technology) through the focussed use of these resources
and environments for a ‘coastal adaptation’to develop (Beaton, 1985;
Thompson and Worth, 2011; Marean, 2014). Hence, while some argue
in terms of a continuum of increasing reliance on coastal resources
and material expression of it, others propose a more operational view
where once a particular socio-economic and technological threshold is
reached resulting in the re-organization of adaptive systems around
coastal resources, a full coastal adaptation is thus involved (Beaton,
1985; Thompson and Worth, 2011; Marean, 2014). Looking for changes
in tool kits to answer these questions might not be productive as many
coastal foraging strategies require simple technology (particularly shell-
fish collection) that often does not preserve well, especially in very early
archaeological contexts (Waselkov, 1987). The comparative study of
shell and faunal records offers more profitable avenues of research as
these records are substantially more abundant than artefacts and are
also reasonably well-preserved. Moreover, well-developed theoretical
frameworks and methodologies can be used to reconstruct procure-
ment and factors behind its variability in space and time (see Bird and
O'Connell, 2006; Lupo, 2007).
When comparing coastal mollusc and faunal records, archaeologists
working with the southern African Late Pleistocene and Holocene
records have looked at relative frequencies or ratios of species and
other proxy measures to evaluate possible marine emphasis in human
diet (e.g., Jerardino, 2010b, 2012; Clark and Kandel, 2013; Dusseldorp
and Langejans, 2013). However, no studies have yet attempted to
meaningfully compare sites on the actual intensity with which re-
sources were procured through time (quantities of preys procured in
time). Intense and systematic use of such resources could well mean
dependency on them and signal a coastal adaptation whichever
definition is followed. The most frequently used proxy-indicators for
identifying recurrent use or dependence on coastal resources (and sev-
eral other types of food and also artefacts) have been density measures
(weight or MNIs or NISP per unit volume) (i.e., Morwood, 1981;
Schweitzer and Wilson, 1982; Sullivan, 1984; Glassow and Wilcoxon,
1988; Thackeray, 1988; Bailey and Craighead, 2003; Langejans et al.,
2012; Faulkner, 2013; Marean, 2014; Will et al., 2015).
1.1. Shell densities and their meaning
A growing body of ethnographic data shows that not all collected
shellfish is transported back to living areas as an important part of
harvested shellfish are processed on the shore (de-shelled or ‘shucked’)
(e.g., Bird and Bliege Bird, 1997; Thomas, 2007). Based on these obser-
vations, one of the obvious archaeological predictions that have been
formulated is that reduced quantities of shells would reach campsites
located at greater foraging distances. Hence, densities of mollusc shells
at archaeological sites should decrease as distance between them and
coastlines increase. Pleistocene records and preliminary observations
among Holocene sites seem to broadly confirm these predictions
(Thackeray, 1988; Smith and Mütti, 2009; Jerardino and Marean,
2010; Dusseldorp and Langejans, 2013). However, reconstruction of
palaeo-shorelines and distances between them and sites have signifi-
cant error margins, either because dating methods probing beyond
the capability of the radiocarbon method have inherently large error
margins (several thousands of years) and/or because bathymetric data
are either coarse or some form of averaging is included in order to ac-
count for foraging radii along several kilometres of coastline that
would have been accessible from a site (e.g., Fisher et al., 2010). Because
sea levels rose and fell very quickly in a matter of a few thousands of
years during interglacial and glacial stages, chronological inaccuracies
can thus translate into flawed foraging distances and biased reconstruc-
tions of the past.
But density values are not without problems. The use of density
values often assumes unchanging matrix composition, constant deposi-
tion rates and adequate preservation of archaeological residues (but see
Parkington, 1988; Jerardino, 1995; Faulkner, 2013). Site matrix can be
generated by people, such as dense shell deposits, or can have a signifi-
cant natural component such as aeolian sand and roof spalling among
others (e.g., Butzer, 1979; Hughes and Lampert, 1982). Hence, variabil-
ity in the densities of shells or fauna may have little to do with the
frequency with which prey was procured and brought back to the site,
but rather depend on the rate of deposition of the dominant component
of matrices which can change significantly over time as revealed by
depth/age curves (i.e., Morwood, 1981; Sullivan, 1984; Stein et al.,
2003).
In order to further explain the problems around the use of shell
density, the following hypothetical cases are presented. For instance,
two shell-bearing sites (A and B) of very similar overall size and depth
of deposit show identical densities of marine shells and fauna, but one
(A) accumulated over a period of time one order of magnitude less
than the other (B). Clearly, the procurement of shellfish and other ma-
rine prey at site A was more intense than at site B, and that, consequent-
ly, marine resources were more systematically procured at site A while
this was not (or much less) the case at site B. An alternative case is that
of two other and similarly sized shell-matrix sites (C and D) with very
different shell densities: site C has densities of marine shells and fauna
one order of magnitude higher than site D, but accumulated over a pe-
riod of one or two orders of magnitude longer than site D. Concluding
that the exploitation of marine resources at site C was more intense
than at site D based on density observationsalone would be most likely
erroneous, becauseshells and fauna accumulated (procured) at site C at
a much slower pace than at site D. Either molluscs were procured with
similar intensity at both sites, and site D received a more substantial
non-anthropic input into its matrix (e.g., aeolian sand) than site C, or
molluscs might well have been utilised more regularly at site D. There-
fore, low-density shell deposits are not synonymous with a distant
shoreline and/or infrequent collection of shellfish.
Similarly, the geological context of sites can also potentially compli-
cate matters, as sites on or near coastal dunes are subject to more sand
input and/or periodical deflation than caves and rock shelters, which
can significantly alter original densities. Weight loss due to shell disso-
lution or burning, and contrasting densities among assemblages with
dissimilar taxonomic composition and/or shell robustness could also
compromise weight-density based comparisonsand minimum number
of individuals (MNIs) density comparisons could thus be more mean-
ingful. Shell densities between assemblages screened with very differ-
ent mesh sizes can also give spurious results, with smaller mesh sizes
retaining greater shell quantities and thus yielding higher densities
(see Jenkins, 2006). Most of these factors have yet to be studied for
their potential impact on observed shell density values.
Given the large error marginsassociated with age determinations of
Pleistocene sites and sometimes low geographic resolution of bathy-
metric data, it is crucial to test the above ethnographically-derived
predictions on shell quantities and transport-mediated decisions with
archaeological observations from shell middens that accumulated at a
time when sea levels were stable. Ideally, shell assemblages ought
to have been sampled with similar field methods from sites located at
differentdistances from the coast, and studied using the same analytical
methods. The best set of data in southern Africa to do this is the mid
and late Holocene record from the West Coast of South Africa. A large
number of shell middens in the Lamberts Bay and Elands Bay coastal
areas have been excavated, extensively radiocarbon dated, their
stratigraphic sequences described, and many of their shell bulk samples
with associated volume information have been analysed and som etimes
fully published. Sea level history and other contextual palaeo-
environmental and geomorphological observations are also available
(Jerardino et al., 2013). Consequently, the objective of this paper is to
approach the issue of systematic use of aquatic resources through
638 A. Jerardino / Journal of Archaeological Science: Reports 6 (2016) 637–644
examining one of the most frequently employed quantitative measures
(proxy-evidence) among archaeologists for inferring such behaviour,
namely shell density. In so doing, the ethnographically-derived predic-
tions of shell density changes with distance from the coast will be tested
with Holocene examples from the South African West Coast. In order to
gain a better understanding of the observed variability, this record is
evaluated against several variables accounted or controlled for such as
screen mesh size, preservation, geomorphological context, taxonomic
composition, and deposition rates. The suitability of shell density and
shell deposition rates for deriving inferences about aquatic resource
procurement and coastal adaptation are then discussed.
2. Methods
The shell assemblages used in this study date to the last 4400 years
and thus after the nearby coastline reached its present position around
8000 years ago (Compton, 2001). These shell assemblages were
excavated following natural stratigraphy, and all were sieved through
a 1/8 in. (3.2 mm) mesh. This is also the case for samples from Pancho's
Kitchen Midden considered here, where although the large majority of
the excavated material was sieved through a 1/16 in. (1.5 mm) mesh
(Jerardino, 1997: 1036), some shell samples were screened with the
above indicated mesh size and thus comparable with the rest of shell
samples used in this study. Shell assemblages are from fourteen sites:
Borrow Pit Midden (BPM), Cape Deseada Midden (CDM), Deurspring
16 (DSP16), Grootrif G (GFG), Grootrif D (GFD), Mike Taylor's
Midden (MTM), Pancho's Kitchen Midden (PKM), Railway Midden
(RWM), Scorpion Shelter (SS), Soutpansklipheuwel 028, 051, and 086
(SPKH028, SPKH051, and SPKH086), Steenbokfontein Cave (SBF), and
Tortoise Cave (TC) (Fig. 1). Observations on the volumes associated
with excavated material, including shell bulk samples, were recorded
during excavations and often ‘bucket note books’were used for this
purpose. Bucket volume observations were later translated into cubic
metres (65 buckets = 1 m
3
). Taxonomic composition and quantifica-
tion (MNIs and weight) were established with detailed laboratory
analyses (for details, see Jerardino, 1997). The measure of taxonomic
Fig. 1. Mapof the South African WestCoast and study area showing locationof sites and places mentioned in the text. CapeDeseada Midden (CDM), Borrow PitMidden (BPM), Deurspring
16 (DSP16),Grootrif G (GFG), Grootrif D (GFD), Mike Taylor's Midden (MTM),Pancho's Kitchen Midden (PKM), Railway Midden(RWM), Soutpansklipheuwel sites (SPKH028, SPKH051,
and SPKH086), Steenbokfontein Cave (SBF), Scorpion Shelter (SS), and Tortoise Cave (TC).
Aerial photographs and maps are from Google Earth and SRTM (v4.1), processed by CGIAR-CSI (http://www.cgiar-csi.org).
639A. Jerardino / Journal of Archaeological Science: Reports 6 (2016) 637–644
composition was expressed as the percentage (%) of one of the most
common mollusc species in local middens, namely, black mussels
(Choromytilus meridionalis). Geomorphological context of these
shell samples was noted as originating from among active or semi-
stabilised and vegetated dunes (0–0.35 km from the coast), and also
in terms of types of site, such as open shell midden, rock shelter or
cave site.
Stratigraphic profiles (section drawings) and available radiocarbon
dates were studied for the purpose of calculating deposition rates. Ideal-
ly, this ought to consider the total volume of material for each period of
occupation under consideration and arrive to measures of total mass or
MNI of fauna (or any other category of archaeological debris for that
matter) per unit of time (see Jerardino, 1995). Because this is often
not possible due to limited information on site surface extent and
variability in middens' depths, a proxy measure of deposition rate is
used instead. Known average depth of sampled midden deposit per
unit of time (i.e., metres/1000 years) is thus employed in this study as
a broad surrogate version of deposition rate. This measure of deposition
rate might not be ideal as the observed thickness of a sampled deposit
may not represent the actual (and perhaps unknown) variability across
sites. Nevertheless, and in the absence of any other better alternative,
meaningful behavioural inferences could be safely made when changes
in deposition rates entail shifts of more than 50% and/or orders of
magnitude. Moreover, in order to gain greater accuracy in the calcula-
tions of deposition rates, these are established for different depositional
events within sites whenever possible.
For the purpose of testing the relationship of shell densities with
average distance from the coast and deposition rates, conventional
regression analyses were performed.
3. Results
Table 1 presents summary observations of shell density data, and
contextual variables, such as distances between sites and nearby coast-
line, type of site, average % black mussels in samples, and deposition
rates as well as observations for calculating the latter. With few excep-
tions, and according to laboratory notes, shell preservation is broadly
similar among sites despite different degrees of exposure to the natural
elements (Table 1). The exceptions are some of SBF samples, where
shells are burnt to different degrees as a result of frequentuse of hearths
(c. 2500–3000 BP) and also post-depositional burning after site aban-
donment (c. 3500–3700 BP) (Jerardino and Yates, 1996: 9). However,
past experimental observations with SBF shells showed that shell
weight loss due to intense burning (i.e., calcination) amounts to about
5% (see Jerardino and Yates, 1996: 10). Despite frequent shell burning,
slightly higher shell weight densities are registered for SBF older
samples that include calcined material (Table 1:SBF[3.5–8.3 ka])
when compared to SBF younger material (SBF [2–3ka]).
Figs. 2 and 3 show a general trend of diminishing mean shell densities
with increasing distance between archaeological sites and coastline. Shell
weight densities drop by half beyond 2 km and MNI densities do so be-
tween 2 and 3 km. Of note is the relatively high variability in mean shell
densities for sites within 0.5 km from the coast, as some of the lowest
and highest values are registered within this initial distance from foraging
localities. Variability in this regard is more evident with weight densities
(Fig. 2). The location of sites showing this higher spread of values corre-
sponds to open sites situated among currently semi-active coastal dunes
and immediately behind them. A linear regression model shows that
this trend is statistically significant (p b0.05) but with a low predictive
power (weight densities: y = −23.382x + 237.18, r
2
= 0.142; MNI
densities: y = −1.786x + 17.484, r
2
= 0.243). Higher coefficients of
determination, and thus better predictive power, are established
when data for the first 0.5 km are not included in the analyses (weight
densities: y = −54.491x + 366.88, r
2
= 0.477; MNI densities:
y=−3.742x + 25.677, r
2
=0.517).
Table 1
Summary observations of shell density data (means and ranges) and associated observations,such as average distances between sites and coastline, type of site, average percentage (%) of black mussels in studied samples (in terms of weight and MNI),
time span during which samples' strata accumulated, average depth of samples' strata, deposition rates of samples' strata and source of these observations.
Site (period represented) Distance from
shore (km)
Type of site % black mussels
(weight)
% black
mussels (MNI)
kg/m
3
(range)
kg/m
3
(average)
MNI/m
3
(range) MNI/m
3
(average)
Time span
(ka)
Average
depth (m)
Deposition
rate (m/ka)
Source of data
DSP16 (2–2.4 ka) 0.17 Open 89.0 66.6 92.2–111.7 101.9 9360–16,336.7 12,848.4 0.4 0.1 0.25 Jerardino (2010b), this study
DSP16 (3–6 ka) 0.17 Open 9.2 5.1 15.1–130.2 55.9 3360.5–23,769.5 10,428.7 2.2 0.9 0.41 Jerardino (2010b), this study
BPM 0.19 Open 68.6 53.5 201.9–474.8 292.3 8753.3–24,093.3 14,756.1 0.1 0.2 2.0 Jerardino et al. (2009), this study
MTM 0.21 Open 89.8 83.3 127.0–517.6 324.2 5297.5–29,575.0 15,683.6 1.3 1.5 1.15 Jerardino and Yates (1997), this study
CDM 0.22 Open 82.0 69.7 123.9–183.3 149.3 13,845.0–22,945.0 19,613.8 0.1 0.1 1.0 Jerardino (2012), this study
GFD 0.40 Open 47.1 23.1 60.9–135.2 101.5 5525.0–20,995.0 12,764.4 0.15 0.5 3.3 Jerardino (2010b), this study
RWM 0.48 Open 96.9 98.0 179.5–354. 266.8 10,205.0–14,235.0 12,220.0 0.1 0.2 1.5 Jerardino (2012), this study
GFG (0–1 ka) 0.60 Open 10.6 8.1 36.8–777.1 341.9 5655.0–21,580.0 34,189.1 1.7 0.5 0.3 Jerardino (2007), this study
PKM (0–1 ka) 1.70 Rock shelter 76.2 52.8 455.9–462.9 459.2 17,712.5–27,941.3 22,826.9 0.3 0.2 0.7 Jerardino (1997), this study
PKM (2–3.6 ka) 1.70 Rock shelter 98.3 86.6 324.8–406.4 365.6 13,422.5–15,000.0 14,211.3 1.15 0.7 0.6 Jerardino (1997), this study
SBF (2–3 ka) 2.60 Cave 83.8 70.0 32.2–429.5 161.3 2292.1–39,325.0 12,639.5 0.8 1.6 2.0 Jerardino (2010b), this study
SBF (3.5–8.3 ka) 2.60 Cave 59.4 40.8 60.1–269.1 192.7 5616.0–20,952.1 19,727.4 4.8 1.8 0.37 Jerardino (2010b), this study
SPKH028 3.60 Open 37.4 25.7 14.7–28.9 21.2 2179.7–2769.2 2486.0 0.5 0.5 1.0 This study
SPKH051 3.60 Open 18.5 12.8 21.5–217.0 124.9 7919.1–10,802.1 9416.6 0.1 0.7 4.0 This study
SPKH086 3.60 Open 48.5 36.1 35.7–59.3 44.1 4407.8–5605.4 5151.5 0.7 0.4 0.5 This study
TC (0–2 ka) 4.00 Rock shelter 82.9 67.9 60.3–263.5 152.8 3770.0–15,560.6 8905.8 1.04 0.45 0.43 This study
TC (3–4.5 ka) 4.00 Rock shelter 80.7 52.1 127.6–217.9 162.4 9425.0–28,470.0 15,042.6 0.7 0.8 0.9 This study
SS 7.40 Rock shelter 94.6 83.1 66.7–93.9 80.7 3052.2–7287.9 5555.6 3.0 0.8 0.27 Wahl (1994), this study
640 A. Jerardino / Journal of Archaeological Science: Reports 6 (2016) 637–644
Except for two or three cases (DSP16 [3–6 ka], GFD and GFG), black
mussel is the most frequent species (≥40%) among studied samples
(Table 1). Intra-site differences in shell densities are evident between
different periods at PKM, SBF and TC, and perhaps in connection with
taxonomic composition: relatively higher values are recorded in sam-
ples with medium percentages of black mussels (47.1–66.6%) whereas
lower shell densities correspond to samples with higher percentages
(66.7–98.3%) of this species (Table 1;Figs. 2 and 3). This is more evident
for MNI density observations (Fig. 3). Otherwise, % black mussels appear
to vary irrespective of shell densities, as high, medium andlow percent-
ages are registered for a range of shell densities (Figs. 2 and 3). In any
case, highest % black mussels are recorded at the most distant site of
all (Table 1;Fig. 1: Scorpion Shelter) situated at 7.4 km from the nearest
accessible rocky reefs.
Shell densities and deposition rates are not statistically related.
Linear regression analyses show very small coefficients of determina-
tion and slopes and are not significantly different from zero (p N0.05)
(weight densities: y = −0.0008x + 1.291, r
2
= 0.008; MNI densities:
y=−0.0236x + 1.475, r
2
= 0.026), indicating no statistical correlation
between shell densities and deposition rates. When data points for sites
within 0.5 km from the coast are excluded from regression analyses, the
relationship between these two variables is still very weak (weight
densities: y = −0.0016x + 1.306, r
2
= 0.039; MNI densities:
y=−0.037x + 1.616, r
2
= 0.091).
4. Discussion
This study shows that, in general, shell densities diminish with
increasing distance between archaeological sites and the coastline. The
observations presented here are controlled for sea level change and
confirm data variability from Pleistocene coastal sites for which
shoreline position has been inferred from bathymetric and palaeo-
shoreline reconstructions (Jerardino and Marean, 2010; Dusseldorp
and Langejans, 2013). However, the geomorphological context of
shell-bearing sites can drive density values to some extent, particularly
at sites situated within 0.5 km from the coast. Substantial aeolian sand
input from dune systems into shell-bearing matrices can lower shell
densities considerably. Hence, initial interpretation of shell densities
for Pleistocene sites for which shoreline position is unknown needs to
be informed by their geomorphological contexts. Low shell densities
could be interpreted as indicative of significant distance from the
coast where shellfish was originally collected, but this might not neces-
sarily be the case. It could well be that molluscs were frequently
harvested within 0.5 km from where shells were finally discarded but
much wind-blown sand from nearby dunes was included into the accu-
mulated shell-bearing deposits, a depositional process that translated
into low shell densities. Although this is evidently pertinent for open
shell-bearing sites that lie unprotected and subject to the action of the
natural elements, substantial input of wind-blown sands can equally
take place in rock shelters and cave sites close to past shorelines
(e.g., Blombos Cave and PP13B cave; Henshilwood et al., 2001; Marean
et al., 2010). Consequently, the geomorphological context of archaeo-
logical sites and reasonably accurate reconstruction of past shorelines
are crucial for interpreting shell densities as a function of distance
from the contemporary coast.
Decreasing shell densities means that people were transporting less
shells to more distant sites. Either less shellfish was taken further away
from the coast altogether, with part or much of the harvest consumed
close to the collection point, or a significant portion of shellfish was
shucked prior to transport back to far-off campsites as shown by ethno-
graphic accounts (Bird and Bliege Bird, 1997; Thomas, 2007). Clearly,
not all consumed or transported shellfish (as flesh or encased in their
shells) is reflected in the quantities of shells found in more remote
shell middens. Consequently, caution must be exercised when inferring
diet/subsistence or settlement scenarios from inter-site comparisons
based on ratios where shell quantities play a role (e.g., mass vertebrate
taxa/mass of marine shells, or number of artefacts/mass of marine
shells).If sites (or occupational episodes within a site) weredifferential-
ly positioned with respect to their contemporary shoreline, ratios may
not be directly comparable and dietary or settlement inferences could
be unwarranted. For instance, the fact that a ratio of mammal bone to
marine shells for a stratigraphic component or site is smaller than that
from another stratum or site associated with a different shoreline loca-
tion does not mean that the collection and/or consumption of molluscs
was relatively more intense in the former relatively to the latter. Shell-
fish harvest intensity and/or consumption might well have been the
same in both instances, with the difference resting on the way similar
quantities of shellfish were processed, transported and consumed.
Hence, knowing the position of former coastlines is key. Comparing
the absolute values of shell-based ratios from sites differentially posi-
tioned from the coast would not be meaningful as these valuesare likely
to be biased by the differential transport of marine shells from the coast
to campsites. However, in the context of a relatively stable shoreline and
similar processing and transport choices, inter- and intra-site compari-
sons of trends through time in shell-based ratios ought to be meaning-
ful. Measuring up the trajectories in such ratios through time among
broadly contemporary sites (whether or not differentially positioned
with respect to the nearest coastline) can inform on parallel or diver-
gent dietary/subsistence trajectories between sites.
Middens with a diverse taxonomic composition are situated at
varying distances from the coast. Among this marked variability,
0
100
200
300
400
500
012345678
Shell Density (kg/ m3)
Average distance from coast (km)
Fig. 2. Plot of shell weight densities as a function of average distances (km) between the
location of sampled sites and the nearest rocky shore. Data point symbols vary according
to average % black mussels in shell samples (triangles: 0–33.3%, squares: 34–66.6%, and
circles: 67–100%).
0
5
10
15
20
25
30
35
40
012345678
Shell Density (MNI/m3) x 1000
Average distance from coast (km)
Fig. 3. Plot of shell MNI densities as a function of average distances (km) between the
location of sampled sites and the nearest rocky shore. Data point symbols vary according
to average % black mussels in shell samples (triangles: 0–33.3%, squares: 34–66.6%, and
circles: 67–100%).
641A. Jerardino / Journal of Archaeological Science: Reports 6 (2016) 637–644
assemblages with the lowest % of black mussel (limpet, and limpet plus
whelk dominated assemblages) are situated closest to the coast, while
those furthest away have some of the highest % black mussels
(e.g., Jerardino, 2007; Wahl, 1994). Although the scope and objective
of this paper is not to explore mollusc inter-site taxonomic variability
in much detail, suffice it to say that demographic and economic trends
that shaped settlement choices, and thus shellfish transport decisions,
explain much but not all of this variability (see Jerardino, 2010b,
2012). In this regard, differences in shell densities between different pe-
riods of time for PKM, SBF and TC deserve some comment. In each of
these instances, higher densities are associated with assemblages with
medium percentages of black mussels (i.e., Jerardino, 1997, 2010b;
Jerardino et al., 2009). These differences are most pronounced with
MNI density observations (Table 1;Fig. 3). Shifts in the mix of prey
and their intrinsically dissimilar shell structure and taphonomy (dis-
tinct three-dimensional shell geometry and specific susceptibilities
to compression) are likely to have played a part in these differences.
Moreover, the relatively larger disparity in PKM shell weight densities
between the pair of periods considered when compared to SBF and TC
equivalents is of note (Table 1;Fig. 2). Large and robust barnacle shells
are abundant in PKM's most recent (N0 AD) shell samples, a factor that
might well explain the highest recorded average shell weight density
among all.
Shell densities have intuitively been interpreted as reflecting a
measure of predation intensity, dietary importance of shellfish in
forager diet and even as an indication of occupation intensity. Doing
so is probably well justified when deposition rates remain unchanged
within single sites or when these are comparable between sites. How-
ever, archaeologists have known for some time that archaeological
deposits of all ages often accumulate in bursts or pulses as a result of
variable and intense human occupation which dictate deposition rates
(e.g., Morwood, 1981; Parkington, 1988, 1990; Stein et al., 2003). More-
over, the concept of predation intensity (intensity of resource use) is in-
extricably linked to dietary returns (in whatever currency is measured)
procured in time (Bird and Bliege Bird, 1997; Lupo, 2007). Consequent-
ly, an archaeological measure for predation intensity ought to factor in
the time dimension, which quantification of shell densities clearly
does not. Hence, the use of shell densities (or of any category of fauna
for that matter) for reconstructing scenarios about shifting foraging
intensities is flawed. Moreover, the empirical data presented here
show in any case that shell densities and deposition rates are not statis-
tically related whether sites within 0.5 m from the coast are considered
or not. Besides, the very weak association between these two variables
is negative and not positive as archaeologists have intuitively assumed
it to be.
The inferences gained from this study are probably applicable
elsewhere; however, it would be beneficial to conduct similar studies
with data from other coastal areas in southern Africa and beyond.
Generalising on the basis of a particular pool of observations for
reconstructing the past in other geographic localities could conceal ad-
ditional local variability. A follow-up study on the variability of shell
densities along the South African south coast is thus desirable, particu-
larly in the context of a large suite of sampled sites available and differ-
ent taxonomic compositions in their shell assemblages (e.g., Klein,
1972; Binneman, 2004/2005; Henshilwood, 2008; Kyriacou, 2009;
Langejans et al., 2012).In this regard, it has been suggested that changes
in the ratios of shell weight to opercula weight for the large sea snail
Turbo sarmaticus is indicative of the removal of the shells (but not oper-
cula) at or near the collecting place for easier transport of the meat to
living sites away from the coast (Henshilwood, 2008; Henshilwood
et al., 2001). The prediction to be tested is that as overall shell densities
decrease with distance from the sea, the weight ratios of T. sarmaticus
shell to opercula must also decrease because these calcareous covers
would become relatively more common in locations more distant
from the coast. Obviously, differential preservation ofshells and opercu-
la as an explanation for higher incidences of the latter in some places
and not in others needs to be ascertained. Moreover, given the different
taxonomic composition and markedly wider range in prey size among
south coast (Indian Ocean) species when compared to their west coast
(Atlantic Ocean) counterpart, it is expected that some of the other
large south coast molluscs (i.e., Haliotis midae,Scutellastra tabularis)
may have been processed more actively and consistently than the
smaller ones. Consequently, their species-specific shell densities may
drop much faster (steeper slopes in plotted data) against distance
from the shore when compared to smaller prey. However, it is also rea-
sonable to expect a measure of variability in their densities as the large
shells of these taxa have been used as containers and as raw materials
for the manufacture of personal ornaments over many thousands of
years (Schweitzer and Wilson, 1982; Henshilwood et al., 2011).
Transporting these large shells back to campsites at varying distances
from the sea would have provided additional benefits, over and above
of their nutritional value, to forager groups.
5. Conclusions
When and where Pleistocene systematic exploitation of marine
coastal resources developed are questions that are often situated within
palaeoenvironmental contexts where coastlines shifted significantly as
a result of glacial and interglacial sea level changes. Some definitions
of systematic exploitation of aquatic resources stress the localised
concentration of mollusc shells as an indicator of it and judgement is
based often on measures such asshell densities. On the other hand, eth-
nographic accounts and preliminary archaeological evidence has shown
that quantities of shells, such as shell densities, vary notably between
campsites and distance to the coast. Other studies have also pointed
out that archaeological densities in general are compromised by various
important factors, such as matrix composition and deposition rates.
Consequently, shell densities needed a closer examination as to their
suitability for inferring past human behaviour. Quantitative approaches
based on shell densities for meaningful inter-assemblage and site com-
parisons have also required re-evaluation as these can be used for
exploring the development of aquatic resource use through time and
across geographic ranges.
This study shows that the use of shell density values can be mis-
leading for inferring intensities of aquatic resource use when rates of
deposition of shell-bearing sites are assumed to be constant and also
when the geomorphological context is not taken into account. The
dependence or not on aquatic resources (or any other for that matter)
is directly related to the intensity with which these are acquired. There-
fore, the time dimension must be factored in when establishing proxy
measures of systematic exploitation of such resources. Shell densities
obviously do not include time as a variable, but deposition rates do so.
Whether distances to contemporary shores are known or not, shell
deposition rates are better proxies for valid comparisons between
sites and through time than shell densities. Ideally, such deposition
rates ought to be quantified in terms of total amount of shells accumu-
lated in a site during a given period of time (see Jerardino, 1995), but
this is not always feasible. Average depth of deposit per unit of time is
presented here as an adequate substitute (see Stein et al., 2003). Differ-
ences in deposition rates by orders of magnitude ought to be meaning-
ful not only in terms of dependence on coastal resources but also for
accommodating relatively broad error margins in age determinations
of Pleistocene deposits and also possible inaccuracies in the assessment
of depth of deposits. Moreover, this study shows that shell densities and
deposition rates are not statistically related, and therefore shell densi-
ties cannot be used as a proxy measure for inferring mollusc foraging
intensity without further observations.
Taxonomic composition and diversity may also affect shell weight
densities and potentially also deposition rates as some species are
more robust than others and susceptible to compression to different
and unknown extents, complicating inter-assemblage comparisons
even further. This may be particularly important when comparing
642 A. Jerardino / Journal of Archaeological Science: Reports 6 (2016) 637–644
assemblages associated with different marine biogeographic zones. It is
thus important to learn more about shell taphonomy among assem-
blages with different species composition and tofollow-upwith studies
from the south coast of South Africa and beyond to improve our under-
standingon the originand evolution of aquatic resource exploitation on
a quantitative basis.
Acknowledgements
I am most grateful to Royden Yates, David Halkett, Tim Hart,
Anthony Manhire, Julia Lee-Thorp and a long list of friends and students
for their help during field work at several of the sites reported here and
discussion of results. Funding of various excavations and the analyses of
samples was provided by grants from the University of Cape Town, the
Centre for Science Development (South Africa), Wenner-Gren (Gr.
5699, Chicago, Illinois) and the Swan Fund (Oxford). Thanks are extend-
ed to Richard Klein and Teresa Steele for stimulating email exchanges
around the topic here addressed and to two anonymous reviewers for
their comments and suggestions. Francesc Conesa kindly compiled
Fig. 1. This contribution would not have been possible without the
archaeo-malacology groundwork set forth in the study area by John
Parkington, Liora Horwitz, Royden Yates, and the late Bill Buchanan. I
am also very grateful to my friends Fiona Archer, René Navarro and
Anette Bestwick for their great and unconditional generosity that has
allowed several data-gathering visits from Barcelona to Cape Town.
CaSEs is an Emergent Grup de Recerca (SGR-1417) supported by the
Generalitat de Catalunya.
References
Avery, G.,Halkett, D., Orton, J., Steele,T., Klein, R.G., 2008.The Ysterfontein 1 MiddleStone
Age rock shelter and the evolution of coastal foraging. S. Afr. Archaeol. Soc. Goodwin
Ser. 10, 66–89.
Bailey, G.N., Craighead, A.S., 2003. Late Pleistocene and Holocene coastal palaeo-
economies: a reco nsideration of th e molluscan evidence from Northern Spain.
Geoarchaeology 18, 175–204.
Bailey, G.N.,Flemming, N.C., 2008.Archaeology of the continentalshelf: marine resources,
submerged landscapes and underwater archaeology. Quat. Sci. Rev. 27, 2153–2165.
Beaton, J.M., 1985. The transition on the coastal fringe of Greater Australia. Antiquity 69,
798–806.
Binneman,J., 2004/2005. Archaeologicalresearch along thesouth-eastern Capecoast. Part
I: Open-air shell middens. S. Afr. Field Archaeol. 13 and 14, 49–77.
Bird, D.W., Bliege Bird, R., 1997. Contemporary shellfish gathering strategies among the
Merriam of the Torres Strait Islands, Australia: testing predictions of a central place
foraging model. J. Archaeol. Sci. 24, 39–63.
Bird, D.W., O'Connell, J.F., 2006. Behavioral ecology and archaeology. J. Archaeol. Res. 14,
143–188.
Butzer, K.W., 1979. Geomorphology and geo-archaeology at Elandsbaai, Western Cape,
South Africa. Catena 6, 157–166.
Clark, J.L., Kand el, A.W., 2013. The evolutionary implications of variation in human
hunting strategies and diet breadth during the Middle Stone Age of Southern
Africa. Curr. Anthropol. 54 (S8), S269–S287.
Colonese, A.C., Mannino, M.A., Bar-Yosef Mayer, D.E., Fa, D.A., Finlayson, J.C., Lubell, D.,
Stiner, M.C., 2011. Marine mollusc exploitation in Mediterranean prehistory: an
overview. Quat. Int. 239, 86–103.
Compton, J.S., 2001. Holocene sea-level fluctuations inferred fro m the evolution of
depositional en vironments of th e southern Langebaan Lagoon salt m arsh, South
Africa. The Holocene 11, 395–405.
Cortés-Sánchez, M.,Morales-Muñiz,A., Simón-Vallejo,M.D., Lozano-Francisco, M.C., Vera-
Peláez, J.L., Finlayson, C., Rodríguez-Vidal, J., Delgado-Huertas, A., Jiménez-Espejo, F.J.,
Martínez-Ruiz, F., Martínez-Aguirre, M.A., Pascual-Granged, A.J., Bergadá-Zapata,
M.M., Gibaja-Bao , J.F., Riquelme-Cantal, J.A., López-Sáez, J.A., Rodrigo -Gámiz, M.,
Sakai, Saburo, Sugisaki, Saiko , Finlayson, G., Fa, D.A., Bicho, N.F., 2011. Earliest
known use of marine resources by Neanderthals. PLoS One 6, e24026.
Dusseldorp, G.L., Langejans, G.H.J.,2013. Carry that weight: coastal foragingand transport
of marine resources during the South African Middle Stone Age. South. Afr. Humanit.
25, 105–135.
Faulkner, P., 2013. Life on the Margins: An Archaeological Investigation of Late Holocene
Economic Variability, Blue Mud Bay, Northern Australia. Australian National Univer-
sity E Press, Canberra.
Fisher, E., Bar-Mathews, M., Jacobs, Z., Jerardino, A., Marean, C.W., 2010. Middle and Late
Pleistocene paleoscape modeling along the southern coast of South Africa. Quat. Sci.
Rev. 29, 1382–1398.
Glassow, M.A., Wilcoxon, L.R., 1988. Coastal adaptations near PointConception, California,
with particular regard to shellfish exploitation. Am. Antiq. 53 (1), 36–51.
Henshilwood, C.S., 2008. Holoce ne prehistory of the southern Cap e, South Africa.
Excavations at Blombos Cave and the Blombosfontein Nature Reserve. Cambridge
Monographs in African Archaeology 75, British Archaeolog ical Reports S1860.
Archaeopress, Oxford.
Henshilwood, C.S., Sealy, J.C., Yates, R., Cruz-Uribe, K., Goldberg, P., Grine, F.E., Klein, R.G.,
Poggenpoel, C., Van Niekerk, K., Watts, I., 2001. Blombos Cave, southern Cape, South
Africa: preliminary report on the 1992–1999 excavations of the Middle Stone Age
levels. J. Archaeol. Sci. 28, 421–448.
Henshilwood, C.S., d'Errico, F., Van Niekerk, K.L., Coquinot, Y., Jacobs, Z., Lauritzen, S.-E.,
Menu, M., García-Moreno, R., 2011. A 100,000-year-old ochre-processing workshop
at Blombos Cave, South Africa. Science 334, 219–222.
Henshilwood, C.S., van Niekerk, K.L., Wu rz, S., Delagnes, A., Armitage, S.J., Rifkin, R.F.,
Douze, K., Keene, P., Haaland, M.M., Reynard, J., Discamps, E., Mienies, S.S., 2014.
Klipdrift Shelter, southern Cape, South Africa: preliminary report on the Howiesons
Poort layers. J. Archaeol. Sci. 45, 284–303.
Hughes, P.J., Lampert, R.J., 1982. Prehistoric population change in southern coastal New
South Wales. In: Bowdler, S. (Ed.), Coastal Archaeology in E astern Australia.
Australian National University Press, Canberra, pp. 16–28.
Jenkins, R.A., 2006. From Midden to Sieve: The Impa ct of Differential Recovery and
Quantification Techniques on Interpretat ions of Shellfish Remains in Australian
Coastal Archaeology BSc (Hons) Dissertation. University of Queensland.
Jerardino, A., 1995. The problem with density values in archaeological analysis: a case
study from Tortoise Cave, Western Cape, South Africa. S. Afr. Archaeol. Bu ll. 50
(161), 21–27.
Jerardino, A., 1997. Changes in shellfish species composition and mean shell size from a
late-Holocene record of the west coast of southern Africa. J. Archaeol. Sci. 24,
1031–1044.
Jerardino, A., 2007 . Excavations at a hunter-gatherer site known as ‘Grootrif G’shell
midden, Lamberts Bay, Western Cape province. S. Afr. Archaeol. Bull. 62 (186),
162–170.
Jerardino, A., 2010a. Prehistoric exploitation of marine resources in southern Africa with
particular reference to shellfish gathering: opportunities and continuities. Pyrenae
41 (1), 7–52.
Jerardino, A., 2010b. Large shell middens in Lamberts Bay, South Africa: a case of hunter-
gatherer resource intensification. J. Archaeol. Sci. 37, 2291–2302.
Jerardino, A., 2012. Large shell middens and hunter-gatherer resource intensification
along the West Coast of SouthAfrica: the Elands Bay casestudy. J. Isl. Coast. Archaeol.
7(1),76–101.
Jerardino, A., Marean, C.W., 2010. Shellfish gathering, marine palaeoecology and modern
human behavior: perspectives from Cave PP13b, Pinnacle Point, South Africa. J. Hum.
Evol. 59, 412–424.
Jerardino, A., Yates, R., 1996. Preliminary results from excavations at Steenbokfontein
Cave: implications for past and future research. S. Afr. Archaeol. Bull. 51 (163), 7–16.
Jerardino, A., Yates, R., 1997. Excavations at Mike Taylor's Midden: a summary report and
implications f or a re-characterization of megamiddens. S. Afr. Archaeol. Bull. 52
(165), 43–51.
Jerardino, A., Dewar, G., Navarro, R., 2009. Opportunistic subsistence strategies among
Late Holocene coastal hunter-gatherers, Elands Bay, South Afr ica. J. Isl. Coa st.
Archaeol. 4, 37–60.
Jerardino, A., Klein, R.G., Navarro, R., Orton, J., Ho rwitz, L.K., 2013. Settlement and
subsistence patterns since the terminal Pleistocene in the Elands Bay and Lamberts
Bay areas. In: Jerardino, A., Malan, A., Braun, D. (Eds.), The Archaeology of the West
Coast of South Africa. Archaeopress, Oxford, pp. 85–108.
Klein, R.G., 1972. Preliminary report on the July through September 1970 excavations at
Nelson Bay Cave, Plettenberg Bay. Palaeoecol. Afr. 6, 177–208.
Kyriacou, K., 2009. The Reinvestigation of Hoffman's/Robberg Cave —The Artefactual and
Shellfish Assemblages M.Phil. Dissertation. University of Cape Town.
Kyriacou, K., Parkington, J.E., Will, M., Kandel, A.W., Conard, N.J., 2015. Middle and Later
Stone Age shellfish exploitation strategies and coastal foraging at Hoedjiespunt and
Lynch Point, Saldanha Bay, South Africa. J. Archaeol. Sci. 57, 197–206.
Langejans, G.H.J., Van Niekerk, K.L., Dusseldorp,G.L.,Thackeray,J.F.,2012.Middle
Stone Age shellfish exploitation: potential indications for mass collecting and re-
source intensification at Blombos Cave and Klasies River, South Africa. Quat. Int.
270, 80–94.
Lupo, K., 2007. E volutionary foraging models in zooarchaeological analysis: recent
applications and future challenges. J. Archaeol. Res. 15, 143–189.
Marean, C.W., 2014. The origins and significance of coastal resource use in Africa and
Western Eurasia. J. Hum. Evol. 77, 17–40.
Marean, C.W., Cleghorn, N., 2003. Large mammal skeletal element transport: applying
Foraging Theory in a complex taphonomic system. J. Taphonomy 1 (1), 15–42.
Marean, C.W., Bar-Matthews, M., Fisher, E., Goldberg, P., Herries, A., Karkanas, P., Nilssen,
P.J., Thompson, E., 2010. The stratigraphy of the Middle Stone Age sedim ents at
Pinnacle Point Cave 13B (Mossel Bay, Western Cape Province, South Africa). J. Hum.
Evol. 59, 234–255.
Morwood, M.J., 1 981. Archaeolog y of the Central Queensland highlan ds: the stone
component. Archaeol. Ocean. 16, 1–52.
Osborn, A., 1977. Strandloopers, mermaids, and other fairy tales: ecological determinants
of marine resource utilization—the Peruvian case. In: Binford, L.R. (Ed.), For Theory
Building in Archaeology. Academic Press, New York, pp. 157–205.
Parkington, J.E., 1988. The Pleistocene/Holocene transition in the western Cape, South
Africa: observation s from Verlorenvlei. In: Bower, J., Lubell, B. (Eds.), Prehistoric
Cultures and Environments in the Late Quaternary of Africa. British Archaeological
Reports International Series 405, Oxford, pp. 197–206.
Parkington, J.E., 1990. A view from the south: southern Africabefore, during and after the
Last Glacial Maximum. In: Gamble,C.S., Soffer, O. (Eds.), The World at 18000 B.P.: The
Low Latitudes vol. 2. Unwin Hyman, London, pp. 214–228.
643A. Jerardino / Journal of Archaeological Science: Reports 6 (2016) 637–644
Parkington, J., 2003. Middens and moderns: shellfishingandtheMiddleStoneAgeofthe
Western Cape, South Africa. S. Afr. J. Sci. 99, 243–247.
Schweitzer, F.R., Wilson, M.L., 1982. Byneskranskop 1, a later Quaternary living site in the
southern Cape Province, South Africa. Ann. S. Afr. Mus. 88, 1–188.
Smith, A.B.,Mütti, B., 2009. A simple measure of intensity of human occupation from shell
density as seen on the Vredenburg Peninsula coast. S. Afr. Archaeol. Bull. 64 (190),
172–175.
Stein, J.K., Deo, J.N., Phillips, L.S., 2003. Big sites —short time: acc umulation rates in
archaeological sites. J. Archaeol. Sci. 30, 297–316.
Sullivan, M.E., 1984. A shell midden excavation at Pambula Lake on the far south coast of
New South Wales. Archaeol. Ocean. 19 (1), 1–15.
Thackeray, J.F., 1988. Molluscan fauna from Klasies River, South Africa. S. Afr. Archaeol.
Bull. 43, 27–32.
Thomas, F.R., 2007. The behavi oral ecology of shellfish gatheri ng in western Kiribati,
Micronesia 1: prey choice. Hum. Ecol. 35, 179–194.
Thompson, V.D., Worth, J.E., 2011. Dwellers by the sea: native American adaptations
along the southern coasts of eastern North America. J. Archaeol. Res. 19, 51–101.
Wahl, E.J., 1994. The Archaeology of Scorpion Shelter B.A. (Hons) Dissertation. University
of Cape Town.
Waselkov, G.A., 1987. Shellfish gathering and shell midden archaeology. Adv. Archaeol.
Method Theory 10, 93–210.
Will, M., Kandel, A.W., Conard, N.J., 2015. Coastal adaptations and settlement systems on
the Cape and Hornof Africa during the MiddleStone Age. In: Conard, N.J.,Delagnes, A.
(Eds.), Settlement Dynamics of the Middle Paleolithic and Middle Stone Age vol. IV.
Kerns Verla g, Tübingen , pp. 61–89.
644 A. Jerardino / Journal of Archaeological Science: Reports 6 (2016) 637–644
