Content uploaded by Philip R Nigst
Author content
All content in this area was uploaded by Philip R Nigst on Jun 05, 2023
Content may be subject to copyright.
Content uploaded by Philip R Nigst
Author content
All content in this area was uploaded by Philip R Nigst on May 20, 2023
Content may be subject to copyright.
ERAUL 150
AnthropologicA et præhistoricA 130
Les sociétés gravettiennes du Nord-Ouest européen:
nouveaux sites, nouvelles données, nouvelles lectures
Gravettian societies in North-western Europe:
new sites, new data, new readings
Actes du colloque international « Le Nord-Ouest européen au Gravettien:
apports des travaux récents à la compréhension des sociétés et de leurs environnements »
(Université de Liège, 12-13 avril 2018)
sous la direction de
Olivier T, Nejma G, Hélène S, Pierre N
Presses Universitaires de Liège
2021
Exploring Diversity of Hunter-Gatherer Behaviour
in the European Mid-Upper Palaeolithic:
e Gravettian Assemblages of Willendorf II and Mitoc-Malu Galben
as Case Studies
Philip R. N*
Marjolein D. B**
Résumé
Exploration de la diversité du comportement des chasseurs-cueilleurs au Paléolithique moyen-supérieur européen:
les assemblages gravettians de Willendorf II et Mitoc-Malu Galben comme études de cas
Cet article explore la variabilité dans les assemblages lithiques et fauniques au Gravettien. En nous appuyant sur l’utilisation
d’in dices et de ratios de diversité d’assemblage, nous examinons quels sont les facteurs importants de variabilité dans la
culture matérielle. Traditionnellement, le Gravettien est divisé en Gravettien ancien, moyen et récent. Ici, nous considérons
le Gravettien dans son ensemble an d’étudier les changements dans la composition lithique et faunique à partir de deux
sites à plusieurs niveaux d’occupation, Willendorf II (Europe centrale) et Mitoc-Malu Galben (Europe orientale). Sur les
deux sites, il n’existe aucune tendance générale importante au cours du temps dans la composition lithique ou faunique,
et ce, que ce soit d’un point de vue strictement chronologique, ce qui pourrait correspondre à des changements dans les
condi tions environnementales, ou que ce soit entre les diérentes sous-phases dénies technologiquement au sein du
Gravettien (i.e. Gravettien ancien, moyen ou récent). Au contraire, les diérences dans la composition lithique ou faunique
sont principalement liées à la localisation du site dans le paysage en termes de, par exemple, la qualité de la matière
première locale ou le type de terrains de chasse à proximité. Les stratégies d’exploitation de la faune sur les deux sites n’ont
pas changé de façon drastique tout au long du Gravettien. De même, il y a peu de changements dans les gisements de
matière première localement disponibles. Nous en concluons donc que les changements observés dans l’outillage lithique
entre le Gravettien ancien, moyen et récent reètent des traditions transmises plutôt que des adaptations fonctionnelles.
Les deux sites ont été le lieu d’activités spéciques et il semble probable que celles-ci ont formé la base des diérences dans
le matériel archéologique retrouvé sur les deux sites. Willendorf II et Mitoc-Malu Galben faisaient probablement partie
de systèmes d’occupation plus larges caractérisés par une grande mobilité et des processus de ssion / fusion au cours des
cycles saisonniers. Le caractère rare et très fragmenté des données archéologiques pour ces chasseurs-cueilleurs gravettiens
sug gère une grande exibilité dans le comportement de ces groupes.
Mots-clés: Paléolithique supérieur, Gravettien, Danube moyen, Est des Carpates, diversité comportementale, exploitation
de la faune, organisation de la technologie lithique.
Abstract
is paper explores variability in Gravettian lithic and faunal assemblages. Using ratios and indices of assemblage diver-
sity we investigate what are the important factors driving variability in material culture. Traditionally the Gravettian is
divided in an Early, Middle and Late Gravettian. Here we consider the Gravettian as a whole to investigate changes in
lithic and faunal composition drawing on two multi-layered sites, namely Willendorf II in Central Europe and Mitoc-
Malu Galben in Eastern Europe. We found that, in our two case-study sites, there are no major trends in either lithic or
faunal composition through time or by technologically dened sub-phases of the Gravettian (i.e., Early, Middle and Late
Gravettian). Instead, our results suggest dierences in lithic and faunal composition are mainly driven by the location of
the site in the landscape in terms of e.g. the quality of local raw material and the type of adjacent hunting grounds. Faunal
exploitation patterns at both sites did not change drastically throughout the Gravettian. Nor did the locally available raw-
material out-crops. We, therefore, propose that the observed changes in lithic toolkits between the Early, Middle, and Late
Gravettian at Willendorf II and Mitoc-Malu Galben reect learned traditions or changes in style rather than functional
* University of Vienna, Department of Prehistoric and Historical Archaeology, Franz-Klein-Gasse 1, Vienna, 1190 (Austria). Email:
philip.nigst@univie.ac.at
** University of Cambridge, McDonald Institute for Archaeological Research, Downing Street, Cambridge, CB2 3ER (United
Kingdom). Email: marjolein.d.bosch@gmail.com
O. Touzé, N. Goutas, H. Salomon & P. Noiret (dir.) (2021) – Les sociétés gravettiennes du Nord-Ouest européen: nouveaux sites, nouvelles données, nouvelles lectures /
Gravettian societies in North-western Europe: new sites, new data, new readings. Liège, Presses Universitaires de Liège
(ERAUL, 150 / Anthropologica et Præhistorica, 130): 287-307
288 P R. N M D. B
adaptations. Both sites were targeted for specic activities and it seems likely that these formed the underpinnings of the
dierences in archaeological remains recovered at both sites. Willendorf II and Mitoc-Malu Galben were likely part of
larger forager settlement systems characterised by high mobility and ssion/fusion processes throughout the seasonal
cycles. e sparse, highly fragmented character of the Gravettian archaeological record suggests these hunter-gatherers
were highly exible foragers.
Keywords: Upper Palaeolithic, Gravettian, Middle Danube, East Carpathians, behavioural diversity, faunal exploitation,
lithic technological organisation.
1. Introduction
Studies on variability in hunter-gatherer behaviour are
oen viewed from chronological and/or geographical
phases (e.g. Svoboda et al., 1996; 1999; Klaric, 2007;
Moreau, 2012). For example, the Gravettian is usually
divided into an Early, Middle, and, Late Gravettian
on the basis of lithic and organic typological markers
such as echettes, gravette and microgravette points,
Kostenki knifes, and organic technology. ere also
exist stone tool industries that are more restricted
in their geographic range or even limited to one or
only a handful of sites, as for example the Maisirian
in Belgium (de Heinzelin, 1973; Campbell, 1980),
the Pavlovian in Moravia (e.g. Svoboda et al., 1996),
or even subdivisions of the Pavlovian (e.g. Polanská
et al., 2014). Nearly all of those taxonomic units are
dened based on typological features (either lithic and
organic tool types or types of reduction sequences),
and might constitute cultural changes between phases
or taxonomic units and as such should inform us
about past hunter-gatherer behaviours. While some
or all of these features are probably related to human
behaviour, their typological nature makes compa-
risons dicult, because the units of analysis are
changing between the named stone tool industries.
Here, we take another approach using quantitative
measures of past hunter-gatherer behaviour focusing
on prey acquisition choices, technological organisa-
tion and landscape use (for similar approaches see,
e.g. Kelly, 1988; Kuhn, 1991; 1995; Stiner et Kuhn,
1992; Roth et Dibble, 1998; Blades, 1999; Blades,
2001; Beck, 2008; Verpoorte, 2009; Surovell, 2012;
Moreau et al., 2016; Clark et Barton, 2017; Barton
et al., 2018; Cascalheira et Bicho, 2018). We employ
a set of diversity measures to quantify taxonomic
heterogeneity and taxonomic dominance in lithic and
faunal datasets as well as ratios of cores, blanks, and
tools to quantify reduction and retouch intensity. A
basic assumption of our study is that prey acquisi tion
choices and landscape use both will have an eect
on lithic technology and lithic assemblage struc ture.
us we should be able to use the latter two to test
hypotheses about prey acquisition choices and land-
scape use.
Hunter-gatherer mobility inuences how humans
(or hominins) organise themselves and their acti-
vities across time and space, as has been shown by
ethnoarchaeological studies in dramatically di erent
environ ments (e.g. Binford, 1980; Kelly, 1983; Kelly,
1992; Grove, 2009; 2010). In this way mobility aects
the organisation of technology and how it struc-
tures the archaeological record. ere are a num-
ber of studies discussing the eects of mobility and
landscape-use on specically lithic tech nological
orga nisa tion and the implications for the archaeo-
logical record (e.g. Binford, 1980; Kuhn, 1991; 1992;
1995; Surovell, 2012; Kuhn et al., 2016).
Lithic technology and prey acquisition techniques
both come with costs and benets. With regard of
lithic technology, the costs are related to procuring,
making, using, and maintaining/resharpening lithic
objects, but also transporting lithics has costs that
might outweigh the benets of transporting them.
ese costs will create variation in lithic archaeological
data sets. Similarly, prey acquisition strategies have
costs and the adapted strategy—regardless if it
involves optimisation in caloric or nutritional return
or any other currency (e.g. secondary products like
antler, hides, etc.)—will shape the composition
of faunal datasets and introduce variation in the
archaeo logical record.
In this study, which we consider as a pilot study
showcasing the approach on a limited dataset, we
conduct our analyses on nine assemblages originating
from two sites that cover the duration of the Gravettian
technocomplex, namely Willendorf II in Central
Europe and Mitoc-Malu Galben in Eastern Europe
(g.1a). e two sites, located in the Middle Danube
region (Willendorf II) and the East Carpathian region
(Mitoc-Malu Galben), were selected because they are
multi-layered sites with high-resolution stratigraphic
and archaeological records and each of them has at
least four archaeological horizons and as such span-
ning a large proportion of the Gravettian as a whole
(g.1b and 1c). Both sites are in bottom slope posi-
tions and similar sedimentary dynamics were at work
during their formation, thus keeping the sedimen tary
context and deposition of the archaeology at the two
sites quite similar as opposed to when comparing, e.g.
E D H-G B E M-U P 289
with cave sites. In addition, from both sites detailed
studies on both fauna and lithics are fully published,
and, thus, make the data necessary for our case
study available and accessible (fauna: enius, 1959;
López Bayón et Gautier, 2007; Noiret, 2009; lithics:
Felgenhauer, 1959; Otte, 1981; 1990; Otte etal., 2007;
Moreau, 2009; 2010; 2012; Noiret, 2009; Moreau
etal., 2016).
Specically, we aim to explore the following ques-
tions:
• In what ways are lithic and faunal diversity
measures dierent between Willendorf II and
Mitoc-Malu Galben?
• Are changes in lithic and fauna diversity related to
relative assemblage age?
• Are lithic and fauna diversity changing between
Early, Middle, and Late Gravettian? If yes, in what
ways are they dierent?
• What are the implications for our understanding
of Late Pleistocene forager land-use and settlement
systems?
Pursuing these research questions requires datasets
on lithic and faunal collections of WillendorfII and
Mitoc-Malu Galben. We used fully published data-
sets (tabl. 1) from both sites and, therefore, focus
this study on the earlier excavations at both sites, as
not all materials from the most recent excavations
(e.g. Nigst et al., 2014; 2021; Noiret et al., 2016) are
available for analysis.
At the outset of this study we should highlight that
the research presented here is part of a larger project
and should be considered a pilot study exploring the
approach on a small dataset. As such, the research has
limita tions that we want to briey mention here—we
discuss them in more detail in the following sections.
One of the biggest limitations is that our nine assem-
blages originate from only two sites. ey might have
been used in similar ways, and, hence, limiting our
sampling of and inferences about Mid-Upper Palaeo-
lithic forager land-use and settlement systems as well
as prey acquisition strategies.
Fig. 1 – a: Map showing the location of Willendorf II and Mitoc-Malu Galben. Base map using GTOPO30
HYDRO1K dataset provided by U.S. Geological Survey’s Center for Earth Resources Observation and Science.
b: Stratigraphic log of the Willendorf II sequence (drawing: P. Haesaerts, modied aer Nigst et al., 2014):
e Gravettian part of the sequence is highlighted.
c: Stratigraphic log of the Mitoc-Malu Galben sequence (drawing: P. Haesaerts, modied aer Haesaerts et al.,
2009): e Gravettian part of the sequence is highlighted.
290 P R. N M D. B
Site AH Culture Relative age
information from Lithic data from Fauna data from
Mitoc-Malu Galben Grav I Middle
Gravettian
Haesaerts, 2007;
Haesaerts et al., 2010 Otte et al., 2007 López Bayón and
Gautier 2007
Mitoc-Malu Galben Grav II Middle
Gravettian
Haesaerts, 2007;
Haesaerts et al., 2010 Otte et al., 2007 López Bayón and
Gautier 2007
Mitoc-Malu Galben Grav III Middle
Gravettian
Haesaerts, 2007;
Haesaerts et al., 2010 Otte et al., 2007 López Bayón and
Gautier 2007
Mitoc-Malu Galben Grav IV Late
Gravettian
Haesaerts, 2007;
Haesaerts et al., 2010 Otte et al., 2007 López Bayón and
Gautier 2007
Willendorf II AH 5 Early
Gravettian
Haesaerts et al., 1996;
Nigst et al., 2014 Otte, 1981 enius, 1959
Willendorf II AH 5+box Early
Gravettian
Haesaerts et al., 1996;
Nigst et al., 2014 Moreau et al., 2016 -
Willendorf II AH 6 Middle
Gravettian
Haesaerts et al., 1996;
Nigst et al., 2014 Otte, 1981 enius, 1959
Willendorf II AH 7 Middle
Gravettian
Haesaerts et al., 1996;
Nigst et al., 2014 Otte, 1981 enius, 1959
Willendorf II AH 8 Middle
Gravettian
Haesaerts et al., 1996;
Nigst et al., 2014 Otte, 1981 enius, 1959
Willendorf II AH 9 Late
Gravettian
Haesaerts et al., 1996;
Nigst et al., 2014 Otte, 1981 enius, 1959
Table 1 – List of assemblages analysed with references. Abbreviations: AH: Archaeological horizon.
2. Sites and Materials
2.1. Site backgrounds
Willendorf II (WII; 48° 19′ 23.50′′ N, 15° 24° 15.20′′E)
is an open-air site located on the le bank of the
Middle Danube river. Embedded in a ~6m deep loess-
palaeosol sequence (g.1b) are at least 11 archaeo-
logical horizons (AH), including Early and Mid-Upper
Palaeo lithic assemblages (Haesaerts et al., 1996; Nigst
et al., 2014). Most of the site was exca vated in several
phases between 1908 and 1955. Later, in the 1980s
and 1990s, eldwork was focused on stratigraphy and
environ mental context of the archaeology (Haesaerts,
1990; Haesaerts et al., 1996). New eldwork has been
undertaken between 2006 and 2011 (Nigst et al.,
2007; 2008a; 2008b; 2008c; 2011a; 2011b; 2012; 2014).
e chronostratigraphic frame work of the site rests
on a detailed stratigraphic sequence and more than
50 radiocarbon dates pro duced on charcoal samples
dated by the Groningen and Oxford radio carbon
labo ratories, placing the sequence between 48 ka and
25 ka BP (∼55–29 ka cal BP) (Nigst et al., 2014).
For this study the upper part of the sequence
inclu ding AH 5 to AH 9, dated to between 30,5 ka and
25ka BP (∼35,5–29 ka cal BP), is of inte rest (g.1b).
e assemblages of AH 5 to AH9 are attributed to
the Mid Upper Palaeolithic, the Gravettian, including
with AH 5 an Early Gravettian, with AH 6 to AH 8
Middle Gravettian, and with AH9 a late Gravettian.
Mitoc-Malu Galben (MMG; 48° 05’ 52” N, 27°
01’ 23′′ E) is also an open-air site with a long loess-
palaeosol sequence. It is situated in the Prut valley
in north-eastern Romania. Within the ~14m deep
sequence (g. 1c), a multitude of archaeological
horizons has been documented (Haesaerts, 2007),
grouped by the initial excavators into 9 layers or
phases. Most of the site was excavated between 1978
and 1990 by V. Chirica with additional eldwork
carried out between 1991 and 1995 by a Romanian-
Belgian team (Chirica, 2001; Otte et al., 2007). More
recently, eldwork at Mitoc-Malu Galben has been
conducted between 2013 and 2016 by a Romanian-
Belgian-British team (Chirica et al., 2014; 2015; 2016;
Noiret et al., 2016; Libois et al., 2017; 2018; Nigst
etal., 2021). e chrono stratigraphy of the sequence
is based on the strati graphic record and more than 40
radiocarbon dates (Haesaerts, 2007).
Here, the Gravettian phases in the upper part
of the sequence are included in our study (g.1c),
namely the phases Grav I to Grav IV. Grav I to
GravIII represent Middle Gravettian, while Grav IV
is classied as Late Gravettian (Otte et al., 2007).
E D H-G B E M-U P 291
2.2. Materials
e assemblages of MMG and WII used in this study
are from the earlier excavations at the two sites.
e ve WII assemblages used here were excavated
between 1908 and 1955. Already during the 1908
and 1909 excavations the materials were curated
by archaeo logical horizon (Felgenhauer, 1959;
Haesaerts et al., 1996; Antl-Weiser, 2008; Nigst, 2012;
Moreau et al., 2016). e four MMG assemblages
utilized in our research were excavated between 1978
and 1995. ey have been mostly recorded by depth
and assigned to stratigraphic units based on the
1991-1995 stratigraphic work (Haesaerts, 2007). We
collected the data from Otte et al. (2007) and used the
archaeo logical layers (i.e., Grav I to Grav IV) dened
by the excavators (Otte et al., 2007). Some of these
layers span more than one stratigraphic unit.
WII AH 5 is attributed to the Early Gravettian
and com prises ~700 lithics and 42 fauna specimens
(tabl.2 and 3). AH 6 to AH 8 are Middle Gravettian
with simi lar lithic and bone recovery. e Late
Gravettian AH 9 has provided ~2580 lithics and
195 fauna speci mens. Detailed descriptions of the
lithic tool types, technological sequences, and fauna
are fully published (fauna: enius, 1959; lithics:
Felgenhauer, 1959; Otte, 1981; Otte, 1990; Moreau,
2009; 2010; 2012; Moreau et al., 2016). In MMG
there is no Early Gravettian present, Layers Grav I to
GravIII are attri buted to the Middle Gravettian. e
lithic data sets of those layers comprise between ~2200
and ~2900 pieces and the Number of Identied Speci-
mens (NISP) of all fauna ranges from 11 to 74. Layer
GravIV is a Late Gravettian with ~11,450 lithics and,
with regard of the fauna, a NISP of 139 (tabl. 2 and3).
e faunal and lithic datasets are fully described and
published (fauna: López Bayón et Gautier, 2007;
Noiret, 2009; lithics: Otte et al., 2007; Noiret, 2009).
Fauna is poorly preserved at both WII and MMG, but
the material was kept and studied in detail.
Assemblage n lithics n cores n blanks n tools
MMG-Grav I 2263 57 2170 36
MMG-Grav II 3756 42 3633 81
MMG-Grav III 2922 67 2814 41
MMG-Grav IV 11469 295 11064 110
WII-AH5 706 23 556 127
WII-AH5+box 2308 66 1970 272
WII-AH6 310 24 221 65
WII-AH7 488 30 318 140
WII-AH8 1425 78 915 432
WII-AH9 2586 75 1715 796
Table 2 – Lithic raw data used in the analysis.
Abbreviations: AH: Archaeological horizon, WII:
Willendorf II, MMG: Mitoc-Malu Galben.
3. Methods
Dierences in taphonomic history between sites and
individual archaeological horizons may have resulted
in dierential faunal preservation which would indi-
cate a post-depositional preservation bias in the fau-
nal material that may have aected lithics as well.
Before comparing assemblages statistically, it rst has
to be established if these potential dierences could
signicantly inuence the results (Grayson, 1984).
To this end we tested the relation between NISP
and Minimum Number of Individuals (MNI) carrying
out a best-t regression analysis between logNISP to
logMNI using Pearson’s correlation coecient (as
the values are not ranked). A signicant correlation
sug gests a similar taphonomic history for all assem-
blages, which, in turn, would warrant comparison.
e approach applied in this study quanties
variation in lithic and faunal datasets by using a
number of diversity indices and ratios. With regard
of faunal exploitation strategies, we employ diversity
indices on the total faunal assemblage as well as on
sub sets such as taxa exploited for subsistence pur-
poses (e.g. horse, bison, rhinoceros, mammoth, red
deer, rein deer, roe deer, and ibex), or for their fur
(e.g. bear, wolf, arctic and red fox and hare). In the
analysis of lithic artefacts, we look at the diversity of
basic lithic categories (cores, blanks, and tools), lithic
reduc tion intensity and retouch intensity.
In our analysis of faunal assemblage composition
and diversity, we use taxonomic heterogeneity
(H; Shannon/Shannon-Wiener index of diversity;
Shannon, 1948; Spellerberg et Fedor, 2003; Magurran,
2004; Lyman, 2008) and taxonomic dominance (1/D;
inverse Simpson index of diversity; Simpson, 1949;
Magurran, 2004; Lyman, 2008). Both indices describe
assem blage composition, but with subtle dierences.
e Shannon index focusses on the evenness in the
distribu tion of all taxa, while inverse Simpson’s index
places emphasis on frequently exploited taxa (Lyman,
2008). e larger the value of H, the greater is the
taxo nomic heterogeneity. e higher the 1/D value
the more evenly species are distributed, whereas low
values signify dominance by one or few species.
Lithic diversity here is used to describe the com-
po si tion of the assemblages with regard of the major,
basic lithic categories, i.e., cores, blanks, and tools
(Glauberman, 2016). As with faunal diversity we use
292 P R. N M D. B
H and 1/D to explore lithic diversity. As with ratios of
reduc tion intensity, these diversity indices quantify
assem blage composition which to a large degree is
related to landscape use and technological organisa-
tion. Core reduction intensity is analysed utilising
blank/core ratio and tool/core ratio, while retouch
inten sity is assessed using tool/blank ratio and tool/
core ratio (e.g. Kuhn, 1991; 1992; 1995; Stiner et
Kuhn, 1992; Blades, 1999; 2001).
All statistical analysis has been conducted in
R3.5.3 (R Core Team, 2019). For bootstrapping we
used the R package boot (Davison et Hinkley, 1997;
Canto et Ripley, 2017), graphics were produced using
the R packages ggplot2 (Wickham, 2016) and cowplot
(Wilke, 2019). Diversity indices were calculated using
the R package vegan (Oksanen et al., 2019).
As test for normality we employed Shapiro-Wilk
test (Royston, 1982a; 1982b; 1995). Correlation
between the used variables and sample size was
tested employing Pearson’s correlation coecient
(Hollander et Wolfe, 1973; Lyman, 2008; Kloke et
Mckean, 2015). As a number of our ratios/indices
were not normally distributed, we used bootstrapping
(Davison and Hinkley, 1997) to calculate non para-
metric condence intervals (adjusted bootstrap
percentile (BCa) intervals (Carpenter et Bithell,
2000)) for Pearson’s correlation coecient. For
testing cor relation between ratios/indices and
ranked data (like the relative chronological position),
we used Spearman’s rank correlation coecient
(Hollander et al., 2014; Kloke et Mckean, 2015). To
compare continuous variables of two samples we
used Wilcoxon (Mann-Whitney U) test and of three
or more sam ples Kruskal-Wallis test (Hollander
et Wolfe, 1973; Kloke et Mckean, 2015). For our
analysis of lithic and faunal diversity indices and
ratios we also employed general linear models with
one or more xed eects (Chambers, 1992; Gelman
et Hill, 2007).
MMG WII
Species Type Grav I Grav II Grav III Grav IV AH5 AH6 AH7 AH8 AH9
Mammuthus primigenius meat 0 5 1 2 2 0 3 1 6
Coelodonta antiquitatis meat 0 0 0 4 0 0 0 0 0
Equus sp. meat 7 23 40 64 0 0 0 4 9
Bos taurus/Bison priscus meat 2 3 24 48 0 0 2 0 2
Capra ibex meat 0 0 0 0 11 2 0 20 51
Ovicaprid meat 0 0 0 0 0 0 0 0 1
Megaloceros giganteus meat 0 0 1 1 0 0 0 0 0
Cervus elaphus meat 0 0 0 0 2 0 0 2 8
Rangifer tarandus meat 2 10 8 20 18 4 4 3 15
Lepus sp. fur 0 0 0 0 3 0 0 1 0
Aves - 0 0 0 0 0 0 0 0 2
Ursus arctos fur 0 0 0 0 2 1 1 0 3
Panthera spelaea fur 0 0 0 0 1 1 0 4 2
Canis lupus fur 0 0 0 0 1 1 1 1 10
Vulpes vulpes fur 0 0 0 0 2 0 1 1 38
Vulpes alopex fur 0 0 0 0 0 0 0 0 46
Gulo gulo fur 0 0 0 0 0 0 0 0 2
MNI total - 5 11 18 30 14 5 8 11 52
NISP total - 11 41 74 139 42 9 12 37 195
NISP fur - 0 0 0 0 9 3 3 7 101
NISP meat - 11 41 74 139 33 6 9 30 92
Table 3 – Faunal raw data used in the analysis. Number of Identied Specimens (NISP) of the individual species
and animals exploited for meat and fur as well as Minimum Number of Individuals (MNI) and NISP for all
species. Abbreviations: AH: Archaeological horizon, WII: Willendorf II, MMG: Mitoc-Malu Galben.
E D H-G B E M-U P 293
Assemblage blank/core
ratio
tool/blank
ratio
tool/core
ratio
H
lithics
1/D
lithics
H
fauna
1/D
fauna
H
meat
1/D
meat
MMG-Grav I 38,070 0,017 0,632 0,199 1,087 0,908 2,123 0,908 2,123
MMG-Grav II 86,500 0,022 1,929 0,165 1,068 1,116 2,535 1,116 2,535
MMG-Grav III 42,000 0,015 0,612 0,183 1,077 1,055 2,442 1,055 2,442
MMG-Grav IV 37,505 0,010 0,373 0,173 1,074 1,202 2,833 1,202 2,833
WII-AH5 24,174 0,228 5,522 0,608 1,530 1,660 3,737 1,037 2,404
WII-AH5+box 29,848 0,138 4,121 0,489 1,345 - - - -
WII-AH6 9,208 0,294 2,708 0,767 1,792 1,427 3,522 0,637 1,800
WII-AH7 10,600 0,440 4,667 0,809 1,958 1,633 4,500 1,061 2,793
WII-AH8 11,731 0,472 5,538 0,805 1,972 1,565 3,049 1,063 2,093
WII-AH9 22,867 0,464 10,613 0,738 1,868 2,019 5,681 1,373 2,810
Table 4 – Shannon (H) and inverse Simpson (1/D) indices and ratios used in the analysis.
Abbreviations: AH: Archaeological horizon, WII: Willendorf II, MMG: Mitoc-Malu Galben.
4. Results and Discussion
Table 4 presents the calculated ratios and diversity
indices for faunal and lithic datasets.
4.1. Exploring potential bias through excavation and
preservation
We identied three possible sources of bias for the
assem blages under study namely a potential lack of
sieving leading to underrepresentation of smaller
specimens, collection bias during excavation/cura-
tion leading to overrepresentation of identiable/
characteristic specimens, and dierence in tapho-
nomic history of the assemblages that might hamper
com parability of (especially faunal) assem blages.
4.1.1. Underrepresentation of small fraction
While WII—compared to MMG—is the older exca-
va tion (the sample used here originates in its vast
majority from the 1908 and 1909 excavations) and,
there fore, oen by default assumed to be missing
smaller elements (size-bias), records in the Natural
History Museum Vienna show that during the 1908
and 1909 excavations sometime sieving was employed,
some times not (Antl-Weiser, 2008). Interestingly,
Moreau (2009) argues based on a compa rison of the
size distribution of projectiles from WII AH5 and
Geissenklösterle AH I, where exca va tions employed
sieving throughout, that the relatively high propor-
tion of small projectiles at WII AH 5 is a good indi-
cator for sieving during the excavation of AH 5. In
sum, this suggests that some of the assemblages are
probably biased by an underrepresentation of smaller
speci mens, but the question is to what degree, as for
example it is unknown whether specic archaeo-
logical horizons were sieved whiles others were not.
On the other hand, neither the 1978-1990 excavations
nor the 1991-1995 excavations at MMG employed
sieving (Otte et al., 2007), and, hence, are both biased
against the small fraction. However, the presence
of some small lithic objects, including fragments of
micro gravettes and backed bladelets, in both the WII
and MMG collections (WII: Felgenhauer, 1959; Otte,
1981; Moreau, 2009; MMG: Otte et al., 2007) suggests
the excavation methods were similarly thorough and
the bias in small material recovery was potentially
comparable for both sites.
4.1.2. Overrepresentation of tools and cores
(or ‘interesting’ pieces)
e WII assemblages of AH 5 to AH 9 described by
Otte (1981) comprise a curated selection of the entire
exca vated material by the original excavators in 1908
and 1909. About twenty years aer Otte’s publica tion,
boxes with additional, typologically less interesting
lithic materials were located in the storage of the
Natural History Museum in Vienna (Nigst, 2004;
2012; Antl-Weiser, 2008; Nigst et al., 2014; Moreau
et al., 2016). Moreau et al. (2016) re-analysed AH 5
including the additional material from the box (here-
aer AH 5+box). To test the extent of curational bias,
we compare all the ratios and indices used in this
paper between the AH 5 and AH 5+box assemblages
(tabl.4 and g.2). ere is a substantial dierence in
the blank/core ratio between AH 5 and AH 5+box,
where the latter has more blanks per core (tabl.4 and
g.2e). e tool/blank ratio (tabl.4 and g.2d) also
shows substantial dierence between AH 5 and AH
5+box, including the lithics from the storage box
results in fewer tools per blank. In both cases the WII
AH 5+box values are in the range of the other WII
values. e same is true for the H and 1/D values
(tabl.4 and g.2a and 2b).
294 P R. N M D. B
Fig. 2 – Comparison of lithic diversity indices (a-b) and ratios (c-e) between the assemblages of Willendorf II
(WII), of Mitoc-Malu Galben (MMG), and the enlarged Willendorf II- AH 5+box assemblage (WII-AH5+box).
Abbreviations: H: Shannon index of diversity, 1/D: inverse Simpson index of diversity.
E D H-G B E M-U P 295
Overall, the patterns described suggest that there
is some bias in the WII assemblages not including the
storage box materials. Because the boxes of AH 6 to
AH 9 are not analysed yet, we need to limit our study
to the assemblages without the storage box materials.
With regard of the tool/core ratio, there is a dif-
ference between AH 5 and AH 5+box, but the AH
5+box value remains within the range of the other
WII tool/core ratios. erefore, despite the fact that
the WII materials used here originate from the old
exca vations and we, in turn, might expect an over-
representation of tools and cores (curational bias),
we can—based on the above described patterns—
assume that the tool/core ratio is least aected by the
curational bias.
4.1.3. Comparability of assemblages using faunal data
Faunal preservation at both sites leaves a lot to be
desired. Dierences in taphonomic history between
sites and individual archaeological horizons may have
resulted in dierential preservation rendering them
unsuitable for comparative purposes. Faunal remains,
which are of course more prone to diagenetic biases
than stone tools, are ideally suited to identify such
di erences in preservation and assess compatibility
of assem blages. e signicant correlation between
logNISP and logMNI across all MMG and WII faunal
assem blages (r=985, t=5.03, df=7, p<0.01) suggests
that they are interrelated in a consistently similar way
and can there fore be compared.
4.2. Sample size eects
In order to test whether ratios and diversity indices
are driven by sample size, we investigated whether
they are correlated to the total number of lithics or
fauna using Pearson’s correlation coecient. e
total number of lithics (n lithics), all lithic diversity
indices and all ratios except the tool/core ratio are
not normally distributed (tabl.5), therefore we used
boot strapping to calculate nonparametric condence
inter vals (adjusted bootstrap percentile (BCa) inter-
vals) for Pearson’s correlation coecient to test the
cor relation between sample size and the ratios and
indices. None of the diversity indices and ratios are
signicantly correlated to sample size (n lithics)
(tabl.6). e diversity indices for overall fauna and
the subset of meat-based exploitation are normally
distributed, while n fauna is not normally distributed
(tabl. 5). Bootstrapped condence intervals for
Pearson’s correlation coecient suggest that none
of the diversity indices are signicantly correlated to
sam ple size (n fauna) (tabl. 6). e above warrants
com parison of assemblages.
variable W p
n lithics 0,710 0,002
H lithics 0,764 0,008
1/D lithics 0,793 0,017
tool/blank ratio 0,809 0,026
blank/core ratio 0,831 0,046
tool/core ratio 0,876 0,143
n fauna 0,809 0,026
H fauna 0,961 0,812
1/D fauna 0,914 0,347
n meat 0,873 0,133
H meat 0,920 0,393
1/D meat 0,921 0,400
Table 5 – Results of tests for normality (Shapiro-
Wilk test [W]). Signicant p-values (<0.05) are
in bold. Abbreviations: H: Shannon index of
diversity, 1/D: inverse Simpson index of diversity.
variable r t df p CI 95% BCa
H lithics -0,612 -2,046 7 0,079 -0.822, 0.181
1/D lithics -0,573 -1,849 7 0,107 -0.776, 0.403
tool/blank ratio -0,533 -1,667 7 0,139 -0.735, 0.685
blank/core ratio 0,378 1,081 7 0,316 -0.538, 0.841
tool/core ratio -0,401 -1,157 7 0,285 -0.739, 0.624
H fauna 0,391 1,122 7 0,299 -0.601, 0.889
1/D fauna 0,466 1,395 7 0,206 -0.630, 0.935
H meat 0,690 2,522 7 0,040 -0.022, 0.836
1/D meat 0,637 2,184 7 0,065 -0.301, 0.923
Table 6 – Results of tests whether Shannon (H) and inverse Simpson (1/D) indices and ratios are correlated
to sample size (Pearson’s Correlation Coecient (r), bootstrapped 95% condence intervals using adjusted
bootstrap percentile intervals [CI 95% BCa]). Abbreviations: df: degrees of freedom, t: t-test statistic value.
296 P R. N M D. B
4.3. Trends through time
We do not expect the behaviour tracked with the
variables analysed here to vary just because of passage
of time, but rather that relative assemblage age—or
increasing assemblage age—represents other trends,
e.g. environ mental and climatic change during the
general climatic downturn towards the Last Glacial
Maximum (especially increased aridity). Under this
assump tion, we would expect responses and adapta-
tions by Pleistocene hunter-gatherers to changing,
i.e. deteriorating, environmental and climatic
condi tions, which can be measured by comparing
behavioural variables to relative assem blage age.
is might include changes in mobility, settlement
systems and landscape use as well as changes in prey
acquisi tion strategies, which are all creating variation
in the archaeo logical record.
To explore whether there are any trends in lithic
and faunal diversity through time, we ordered the
assem blages according to their chronostratigraphic
position (WII: Haesaerts et al., 1996; Nigst et al.,
2014; MMG: Otte et al., 2007), both within the indi-
vidual sites as well as between the two sites. For the
latter we used the chronostratigraphic correlation for
the Middle and Late Pleniglacial aer Haesaerts et al.
(2004; 2009; 2010).
Analysis including both WII and MMG lithic
assem blages does not show any trend in time, nor
do investigations of the sites separately (tabl. 7).
Equally, none of the faunal diversity indices show any
chrono logical trend, with the exception of ani mals
that were primarily exploited for their meat. In the
latter instance, H but not 1/D is correlated with rela-
tive chronological position of our entire data set, i.e.,
the nine Gravettian assemblages of WII and MMG.
is might indicate that over time the composition of
prey taxa primarily hunted for subsistence purposes
became more even (less specialised), but the distribu-
tion of frequently exploited taxa (i.e., the number of
preferred prey species) did not signicantly change.
is change in species composition may point to a
weak increase in exploitation pressure, which could
be caused by the deteriorating climatic conditions
towards the LGM, or simply signify a shi towards
a moderately more opportunistic hunting strategy.
How ever, it is more likely that the shi in meat
exploitation represents behavioural variability rather
than an adaptation to the climatic downturn towards
the LGM, because the taxonomic evenness and
dominance of the total faunal spectrum (meat and fur
exploita tion) do not change signicantly over time.
Variable WII + MMG WII MMG
SpSpSp
H lithics 156 0,437 14 0,683 14 0,750
1/D lithics 140 0,678 6 0,233 14 0,750
tool/blank ratio 140 0,678 2 0,083 18 0,333
blank/core
ratio 88 0,493 20 1,000 14 0,750
tool/core ratio 144 0,613 6 0,233 18 0,333
H fauna 126 0,912 14 0,683 2 0,333
1/D fauna 128 0,880 14 0,683 2 0,333
H meat 22 0,011 2 0,083 2 0,333
1/D meat 40 0,059 10 0,450 2 0,333
Table 7 – Results of Spearman’s rank order correlation (S) of Willendorf II (WII) and Mitoc-Malu Galben
(MMG) to test whether Shannon (H) and inverse Simpson (1/D) indices and ratios are related to relative
assemblage age. Signicant p-values (<0.05) are in bold.
4.4. Changes between Early, Middle, and Late
Gravettian
We also explored whether we can identify any signi-
cant dierences in the studied variables between
Early, Middle and Late Gravettian assemblages. ere
could be cultural dierences, changes in landscape-
use, mobility and/or the duration of site occupation.
ese typologically grouped comparisons are
hampered by the fact that there is only one Early
Gravettian assemblage and only two Late Gravettian
assem blages in our dataset. As a result, no site based
analyses could not be conducted. However, analysis
of WII and MMG assemblages together suggests no
signi cant dierences between Early, Middle, and
Late Gravettian (tabl. 8 and g. 3). is suggests
E D H-G B E M-U P 297
that faunal exploitation did not change between
the Gravettian phases, nor did lithic reduction or
retouch intensity. One could argue that at WII and
MMG there are no apparent changes in landscape-
use and, therefore, we propose that the changes in
tool kit morpho logy, on which the Gravettian phases
are based, could reect dierent learned traditions.
In other words, changes in lithic toolkit morphology
and the related production technology signify inno-
va tions providing new solutions to existing tasks
rather than adaptations to changing conditions. is
will have to be evaluated against a larger dataset in
the future.
variable Chi2df p
H lithics 0,356 2 0,837
1/D lithics 0,089 2 0,957
tool/blank ratio 0,089 2 0,957
blank/core ratio 0,000 2 1,000
tool/core ratio 0,622 2 0,733
H fauna 2,600 2 0,273
1/D fauna 1,689 2 0,430
H meat 4,356 2 0,113
1/D meat 4,200 2 0,123
Table 8 – Results of Kruskal-Wallis test to
compare Shannon (H) and inverse Simpson (1/D)
indices and ratios between Early, Middle and Late
Gravettian. Signicant p-values (<0.05) are in
bold. Abbreviations: df: degrees of freedom.
4.5. Dierences between sites
ere are substantial dierences in the lithic indices
and ratios between WII and MMG assemblages. As
shown in Table 9 and Figures 4a to 4e, comparison
between the MMG and WII assemblages shows
signi cantly higher diversity in all WII assemblages.
More over, all diversity indices show a broader range
of variation among WII assemblages, whereas the
studied MMG assemblages cluster tighter together.
e more evenly distributed lithic assemblages at
WII may relate to lower blank frequencies, which in
turn might reect a curational bias (see discussion
above), and/or relate to lower rates of retouched
pieces at MMG.
e tool/blank and tool/core ratios are signicantly
higher at WII, suggesting higher rates of retouch,
both when comparing the number of retouched
pieces to blanks and to cores, than at MMG (tabl.9
and g.4c and 4d). Conversely, the blank/core ratio
is signicantly higher at MMG (tabl.9 and g.4e),
which indicates that reduction intensity of cores,
i.e. the number of blanks produced from one core
is signicantly higher at MMG. e latter may be
explained by the proximity of MMG to high-quality
raw-material, whereas a large portion of raw-material
at WII constitutes local, low-quality Danube gravels.
In terms of fauna diversity, H and 1/D show
more specialised faunal exploitation, i.e., dominated
by fewer taxa (tabl.9 and g.4f and 4g), at MMG.
WII, which exhibits more even faunal exploitation,
is situated in the Wachau valley where the Danube
cuts through the foothills of the Bohemian Massif.
e small river plain is surrounded by rocky, in part
forested slopes. On the contrary, MMG lies on the
at to hilly but open plain of the Prut. In fact, fau-
nal exploitation patterns at both sites are clearly
embed ded in the habitats surrounding the sites. At
MMG steppe animals such as horse and to a lesser
extent bison were primarily targeted (López Bayón
et Gautier, 2007; Noiret, 2009), while at WII there
is a wider range of rocky terrain (e.g. ibex), forest
(e.g. red deer, fox, glutton), and oodplain or more
open terrain (e.g. mammoth, horse) species (enius,
1959).
Contrary to the above mentioned trend through
time to broader dietary faunal exploitation (H meat),
the diversity of species exploited for meat does not
dier between the two sites (tabl.9 and g.4h andi).
Hence, the signicant dierence in overall faunal
com position rests on secondary (e.g. fur, ivory,
antler, etc.) rather than primary exploitation for meat
consump tion.
variable W p
H lithics 20 0,016
1/D lithics 20 0,016
tool/blank ratio 20 0,016
blank/core ratio 0 0,016
tool/core ratio 20 0,016
H fauna 20 0,016
1/D fauna 20 0,016
H meat 9 0,905
1/D meat 7 0,556
Table 9 – Results of Wilcoxon test (W) to compare
Shannon (H) and inverse Simpson (1/D) indices
and ratios between Willendorf II and Mitoc-Malu
Galben. Signicant p-values (<0.05) are in bold.
298 P R. N M D. B
Fig. 3 – Comparison of diversity indices and ratios between Early, Middle, and Late Gravettian.
Abbreviations: H: Shannon index of diversity, 1/D: inverse Simpson index of diversity, Grav: Gravettian.
E D H-G B E M-U P 299
Fig. 4 – Comparison of diversity indices and ratios between the assemblages of Willendorf II (WII) and Mitoc-
Malu Galben (MMG). Abbreviations: H: Shannon index of diversity, 1/D: inverse Simpson index of diversity.
300 P R. N M D. B
Fig. 5 – Scatterplots showing the correlation of lithic diversity indices/ratios and faunal diversity.
Green: Mitoc-Malu Galben assemblages. Red: Willendorf II assemblages. Regression lines based on linear
regression are shown in blue, 95% condence intervals in grey.
Abbreviations: H: Shannon index of diversity, 1/D: inverse Simpson index of diversity.
E D H-G B E M-U P 301
response ~ xed eect 1 response ~ xed eect 1 + xed eect 2
response
variable xed eect 1 xed eect 2 lm(response ~ xed eect 1) lm(response ~ xed eect 1 + xed
eect 2)
correlation xed
eect 1
correlation xed
eect 2
H lithics H fauna site adj.R2=0.654, F(1,7)=16.1, p=0.005 adj.R2=953, F(2,6)=82.43, p<0.001 t=-0.619, p=0.559 t=-6.766, p=0.001
1/D lithics 1/D fauna site adj.R2=0.479, F(1,7)=8.363, p=0.023 adj.R2=0.876, F(2,6)=29.21, p<0.001 t=0.285, p=0.785 t=-4.832, p=0.003
tool/core ratio H fauna site adj.R2=0.875, F(1,7)=57, p<0.001 adj.R2=0.868, F(2,6)=27.29, p<0.001 t=4.257, p=0.005 t=0.792, p=0.459
tool/blank ratio H fauna site adj.R2=0.732, F(1,7)=22.9, p=0.002 adj.R2=0.833, F(2,6)=20.89, p=0.002 t=0.945, p=0.381 t=-2.277, p=0.063
blank/core ratio H fauna - adj.R2=0.236, F(1,7)=3.475, p=0.105 - - -
tool/core ratio 1/D fauna site adj.R2=0.760, F(1,7)=26.35, p=0.001 adj.R2=0.756, F(2,6)=13.4, p=0.006 t=2.658, p=0.038 t=-0.940, p=0.383
tool/blank ratio 1/D fauna site adj.R2=0.554, F(1,7)=10.95, p=0.013 adj.R2=0.831, F(2,6)=20.61, p=0.002 t=0.902, p=0.402 t=-3.523, p=0.013
blank/core ratio 1/D fauna - adj.R2=0.147, F(1,7)=2.381, p=0.167 - - -
H lithics H meat - adj.R2=-0.127, F(1,7)=0.100, p=0.761 - - -
1/D lithics 1/D meat - adj.R2=-0.129, F(1,7)=0.087, p=0.777 - - -
tool/core ratio H meat - adj.R2=0.095, F(1,7)=1.837, p=0.218 - - -
tool/blank ratio H meat - adj.R2=-0.129, F(1,7)=0.086, p=0.778 - - -
blank/core ratio H meat - adj.R2=-0.071, F(1,7)=0.467, p=0.517 - - -
tool/core ratio 1/D meat - adj.R2=-0.078, F(1,7)=0.420, p=0.538 - - -
tool/blank ratio 1/D meat - adj.R2=-0.143, F(1,7)=0.001, p=0.984 - - -
blank/core ratio 1/D meat - adj.R2=-0.081, F(1,7)=0.402, p=0.546 - - -
Table 10 – Linear Models to study Shannon (H) and inverse Simpson (1/D) indices of lithic diversity and lithic ratios as a function of Shannon (H) and inverse Simpson
(1/D) indices of fauna diversity [lm(response ~ xed eect 1] and as a function of fauna diversity plus site [lm(response ~ xed eect 1 + xed eect 2].
Signicant p-values (<0.05) are in bold.
302 P R. N M D. B
4.6. Comparison of lithic and faunal diversity indices
and ratios
Up to now we have analysed lithic and faunal diversity
indices and ratios separately, and we have shown that
most of them are not driven by relative assemblage
age or cultural attribution (i.e., Early, Middle and
Late Gravettian), but seem to be inuenced primarily
by dierences between the two case-study sites.
Below, we compare lithic and faunal diversity
indices with each other because one would expect that
prey acquisition (measured here through composition
and diversity of the faunal assemblages) inuences
lithic assemblage composition and diversity. Equally,
both may be driven by other factors such as climate
or the site’s role in a forager settlement system.
To explore what drives lithic variability we used
linear models with one or two xed eects (tabl.10
and g.5). We constructed a linear model of H lithics
as a function of H fauna. is model was signicant
(F (1,7) =16.1, p=0.005) (tabl. 10 and g. 5a). To
evaluate the impact of site on our model, to account
for the signicant dierences when comparing the
diver sity indices between WII and MMG, we included
it as a second xed eect. e resulting model was
again signicant (F (2,6) =82.43, p<0.001), while
the coecients for each eect clearly show that the
model is driven by site (p=0.001) rather than H fauna
(p=0.559) (tabl.10). e same pattern emerges when
testing the correlation of 1/D lithics to 1/D fauna, as
well as the correlation of tool/blank ratio to faunal
diver sity (both H and 1/D) (tabl.10 and g.5b, 5e,
and 5f).
e two linear models for blank/core ratio as a
func tion of either H or 1/D of fauna are not signicant
(tabl. 10 and g. 5g and 5h). However, when we
investigate the tool/core ratio as a function of either H
or 1/D of fauna, both models are signicant (tabl.10
and g.5c and 5d). When site is added as a second
xed eect the latter two models are still signicant
and driven by faunal diversity. Interesting in this
context is also our suggestion that tool/core ratio
is among the least biased in old collections as tools
and cores were collected in larger percentages even
in old excavations, which normally are biased against
blanks (especially akes and cortical elements) as
well as smaller lithic fractions. e patterns described
for the tool/core ratio suggest that the more even the
faunal assemblage is, the more tools are produced
per core, i.e., the more curated or more reduced the
assem blage is.
While it is possible that the faunal composition
drives the tool/core ratio, it is also possible that both
are connected to other hunter-gatherer behaviours
that inuence both variables, which do not dier
signi cantly between the two sites. For example,
longer occupation duration and re-occupation or
palimpsest of occupations, both leading to more tools
per core and a more even faunal distribution. Another
option is highly residentially mobile groups also lead-
ing to more curated lithic assemblages (many tools
and few cores, partly as transport of cores has higher
costs) and more even assemblages.
Overall, our analysis of the nine Gravettian
assem blages from WII and MMG suggests that site
is a driving factor in most the ratios and indices.
Our variable ‘site’ captures the location of the site
in the landscape, e.g. in terms of access to high vs.
low quality lithic raw materials and access to animal
resources through the type of adjacent hunting
grounds, and its role in the forager settlement system.
5. Concluding remarks
is paper investigated the underpinnings of diversity
in material culture of the Gravettian at Willendorf II
and Mitoc-Malu Galben. Specically, we explored
the dierences in lithic and faunal composition
both from a temporal and cultural perspective and
discuss implications for our understanding of Late
Pleistocene forager land-use and settlement systems.
While the individual variables used in this study
are not driven by sample size, we are aware that due
to the focus on nine assemblages we have a rather
low number of assemblages in our analyses. Also the
fact that these nine assemblages come from only two
case study sites might bias our results. Future studies
of the questions raised here will have to be pursued
including more assemblages from more sites, and will
need to evaluate the reproducibility of the patterns
described.
MMG is close to high-quality raw-material
resources and as a result the blank to core ratio is
higher than at WII where high-quality raw material
out crops are far away (min. 80km) and a large
part of the raw material constitute locally available
(secondary source < 1km distance from site), rather
low-quality Danube gravels. Furthermore, WII is
located on the west bank of the Danube overseeing
the narrow river plain and is surrounded by rugged,
mountainous terrain, which explains the focus on
ibex exploitation and the frequency of small and
larger carnivores (enius, 1959). MMG in the Prut
valley on the at to hilly but open east Carpathian
plain facilitated hunting of horse and bison (Noiret,
2009).
E D H-G B E M-U P 303
rough linear modelling we assessed how faunal
diver sity inuences lithic diversity. Faunal diversity is
signi cantly positively correlated with the tool/core
ratio. In other words, the more even or heterogenic
the faunal composition is, the more tools were pro-
duced per core. Or the fewer tools were made per
core, the more selective the faunal assemblage.
Coming back to our case-study sites, MMG can
from a faunal point of view be characterised as short-
term (1-6 weeks) occupations in which few activities
(e.g. organic tool production [from reindeer antlers],
butchery of large mammals, specically horses and
bison killed close to the site) repeatedly took place
(López Bayón et Gautier, 2007; Noiret, 2009). More-
over, from the lithics perspective MMG can be seen
as a residential camp close to a high-quality raw
mate rial outcrop with high levels of blank production
and core reduction, but only a few tools per core or
per blank. For WII we know little about the duration
of site-occupation although Felgenhauer (1959)
men tions the presence of lenses dense in material
remains. Nevertheless, there is a broader set of acti-
vities recorded at WII, namely lithic knapping,
exploita tion for subsistence purposes of medium-
sizes ungulates e.g. ibex and reindeer, but also larger
mam mals including mammoth, organic tool pro-
duc tion (antler and perhaps ivory), production of
gurative art (ivory and stone), and exploitation of
furbearing animals from hare to foxes, wolves, and
bears (enius, 1959).
Clearly, MMG and WII were targeted for specic
acti vities and it seems likely that these formed the
under pinnings of the dierences in the archaeological
remains recovered at both sites. e multitude of
acti vities conducted at WII, which are reected in the
fau nal composition, may well explain the lithic diver-
sity.
In sum, with the archaeological assemblages of
WII and MMG we probably have repeated samples of
forager settlement systems that are characterised by
high mobility and probably ssion/fusion processes
throughout the seasonal cycles at two dierent loca-
tions, both in terms of season and space. e sparse,
highly fragmented character of the Gravettian
archaeo logical record of these hunter-gatherers sug-
gests highly exible foragers exploiting their land-
scapes to the full.
Acknowledgements
Many thanks to Pierre Noiret, Olivier Touzé, Hélène
Salomon, and Nejma Goutas for the invitation to
the workshop in April 2018 and to contribute to the
publication. Many thanks to Alice Leplongeon for
trans lating the Résumé. We also want to express our
gratitude to two reviewers for their detailed com-
ments, which signicantly improved our manuscript.
MDB’s research was supported by the EC Horizon
2020 Marie Skłodowska-Curie program (EU-BEADS
project, grant no. 656325). PRN’s research was funded
by the EC FP7 Marie Curie program (NEMO-ADAP
project, grant no. 322261), the Leakey Foundation,
the D.M. McDonald Grants and Awards Fund, the
Isaac Newton Trust, and the British Academy (British
Academy/Leverhulme Small Grant).
Author contributions: Both authors designed the
research, collected and analysed data, and wrote the
paper.
Bibliography
A-W W. (2008) – Die Frau von W. Die Venus
von Willendorf, ihre Zeit und die Geschichte(n) um ihre
Au ndung. Vienna, Phoibos Verlag, 208 p.
B C.M., A T J.E., G-PO.,
R-S J.G., G N., V
C M., P B G. (2018) – Risk and
Resilience in the Late Glacial: A Case Study from the
Western Mediterranean. Quaternary Science Reviews,
184, p.68-84.
B R.K. (2008) – Transport Distance and Debitage
Assem blage Diversity: an Application of the Field Pro-
ces sing Model to Southern Utah Toolstone Procure-
ment Sites. American Antiquity, 73 (4), p.759-780.
B L.R. (1980) – Willow Smoke and Dogs’ Tails:
Hunter-Gatherer Settlement Systems and Archaeo-
logical Site Formation. American Antiquity, 45 (1),
p.4-20.
B B.S. (1999) – Aurignacian Lithic Economy and
Early Modern Human Mobility: New Perspectives
from Classic Sites in the Vézère Valley of France. Jour-
nal of Human Evolution, 37 (1), p.91-120.
— (2001) – Aurignacian Lithic Economy. Ecological Pers-
pec tives from Southwestern France. New York, Kluwer
Academic/Plenum Publishers.
C J.B. (1980) – Le problème des subdivisions du
Paléo lithique supérieur britannique dans son cadre
européen. Bulletin de la Société royale belge d’Anthro-
pologie et de Préhistoire, 91, p.39-77.
304 P R. N M D. B
C A., R B. (2017) – boot: Bootstrap R (S-Plus)
Func tions, R package version 1.3-20.
C J., B J. (2000) – Bootstrap Condence
Inter vals: When, Which, What? A Practical Guide for
Medical Statisticians. Statistics in Medicine, 19 (9),
p.1141-1164.
C J., B N. (2018) – Testing the Impact
of Environmental Change on HunterGatherer Settle-
ment Organization during the Upper Paleolithic in
Western Iberia. Journal of Quaternary Science, 33 (3),
p.323-334.
C J.M. (1992) – Linear Models. In: J.M.Chambers,
T.J. Hastie (ed.), Statistical Methods in S. Boca Raton.
London, New York, Washington DC, Chapman &
Hall/CRC, p.95-144.
C V. (2001) – Gisements paléolithiques de Mitoc. Le
paléo lithique supérieur de Roumanie à la lumière des
décou vertes de Mitoc. Iaşi, Helios, 216 p.
C V., N P., N P.R., H P. (2014)
– Mitoc-Malu Galben, Campania 2013. Cronica
cercetărilor arheologica din România, XLVIII, p.81-83.
— (2015) – Mitoc-Malu Galben, Campania 2014. Cronica
cercetărilor arheologica din România, XLIX, p. 153-156.
C V., N P., N P.R., P.,
V D.-M., B G., B M.D., P K.,
G F., L T., B A., B T.,
C D. (2016) – Mitoc-Malu Galben, Campania
2015. Cronica cercetărilor arheologica din România, L,
p.62-63.
C G.A., B C.M. (2017) – Lithics, Landscapes &
la Longue-durée – Curation & Expediency as Expres-
sions of Forager Mobility. Quaternary International,
450, p.137-149.
D A.C., H D.V. (1997) – Bootstrap Methods
and their Application. Cambridge, Cambridge Univer-
sity Press, 582 p.
D H J. (1973) – L’Industrie du site paléolithique
de Maisiéres-Canal. Bruxelles, Institut royal des
Sciences Naturelles de Belgique (Mémoires, 171),
p.3-62.
F F. (1959) – Willendorf in der Wachau.
Mono graphie der Paläolith-Fundstellen I-VII. Vienna,
Rohrer, 420 p.
G A., H J. (2007) – Data Analysis using Regression
and Multilevel/Hierarchical Models. Cambridge,
Cambridge University Press, 625 p.
G P.J. (2016) – Middle Palaeolithic Land Use
in Dutch and Belgian Limburg: Integrating Data from
Sur face Sites. Quaternary International, 411, p. 198-
215.
G D.K. (1984) – Quantitative Zooarchaeology:
Topics in the Analysis of Archaeological Faunas.
Orlando, Academic Press, 202 p.
G M. (2009) – Hunter–Gatherer Movement Patterns:
Causes and Constraints. Journal of Anthropological
Archaeology, 28 (2), p.222-233.
— (2010) – Logistical Mobility Reduces Subsistence Risk
in Hunting Economies. Journal of Archaeological
Science, 37 (8), p.1913-1921.
H P., B I., C V.P., C V.,
D N.I., K L., O L.A.,
V P J., D F. (2010) – Charcoal
and Wood Remains for Radiocarbon Dating Upper
Pleistocene Loess Sequences in Eastern Europe and
Central Siberia. Palaeogeography, Palaeoclimatology,
Palaeoecology, 291 (1-2), p.106-127.
H P., B I., C V., D F.,
K L. (2004) – Cadre stratigraphique et
chrono logique du Gravettien en Europe Centrale. In:
J.Svoboda, L. Sedlácková (ed.), e Gravettian along the
Danube, Proceedings of the Mikulov Conference (20-
21 November, 2002). Brno, Institute of Archaeology,
AS CR, p.33-56.
H P., D F., B M., T G.
(1996) – Revised Stratigraphy and Chronology of the
Willendorf II Sequence, Lower Austria. Archaeologia
Austriaca, 80, p.25-42.
H P., B I., C V.P., C V.,
D F., D N.I., O L.A., PS.,
V P J. (2009) – Climatic Signature and
Radio carbon Chronology of Middle and Late Pleni-
glacial Loess from Eurasia: Comparison with the
Marine and Greenland Records. Radiocarbon, 51 (1),
p.301-318.
H P. (1990) – Nouvelles recherches au gisement
de Willendorf (Basse Autriche). Bulletin de l’Institut
Royal des Sciences Naturelles de Belgique, Sciences de la
Terre, 60, p.203-218.
E D H-G B E M-U P 305
— (2007) – Mitoc-Malu Galben: cadre stratigraphique et
chrono logique. In: M. Otte, V. Chirica, P. Haesaerts
(dir.), L’Aurignacien et le Gravettien de Mitoc-Malu
Galben (Moldavie Roumaine). Liège, Université de
Liège (ERAUL, 72), p.15-41.
H M., W D.A., C E. (2014) – Non-
parametric Statistical Methods. Hoboken, John Wiley
& Sons, 828 p.
H M., W D.A. (1973) – Nonparametric Sta-
tis tical Methods. New York, John Wiley & Sons, 503p.
K R.L. (1983) – Hunter-Gatherer Mobility Strategies.
Jour nal of Anthropological Research, 39 (3), p.277-306.
— (1988) – e ree Sides of a Biface. American Antiquity,
53 (4), p. 717-734.
— (1992) – Mobility/Sedentism: Concepts, Archaeological
Measures, and Eects. Annual Review of Anthropology,
21, p.43-66.
K L. (2007) – Regional Groups in the European
Middle Gravettian: a Reconsideration of the Rayssian
Technology. Antiquity, 81, p.176-190.
K J., MK J.W. (2015) – Nonparametric Statistical
Methods Using R. Boca Raton, CRC Press, 287 p.
K S.L. (1991) – “Unpacking” Reduction: Lithic Raw
Mate rial Economy in the Mousterian of West-central
Italy. Journal of Anthropological Archaeology, 10 (1),
p.76-106.
— (1992) – On Planning and Curated Technologies in
the Middle Paleolithic. Journal of Anthropological
Research, 48 (3), p.185-214.
— (1995) – Mousterian Lithic Technology. An Ecological
Pers pective. Princeton, Princeton University Press,
209p.
K S.L., R D.A., C A.E. (2016) – What
Moves Us? How Mobility and Movement are at the
Center of Human Evolution. Evolutionary Anthropo-
logy, 25 (3), p.86-97.
L T., B G., N P.R., Noiret P. (2017) – Mitoc-
Malu Galben (Roumanie) : quelques aspects de la
techno logie lithique Gravettienne. In: V. Chirica,
C.Ichim (dir.), Les Gravettiens: leur création matérielle
et spiri tuelle. Iasi, Editura Pim, p.78-105.
L T., N P.R., H P., B M.D.,
M W.C., B T., N P. (2018)
– 40 ans des recherches avec Vasile Chirica: Mitoc-
Malu Galben. In: C.M. Lazarovici, A. Berzovan
(dir.), Quaestiones Praehistoricae. Studia in honorem
Professoris Vasile Chirica. Bucarest and Brăila, Editura
Academiei Române – Editura Istros, p.53-74.
L B I., G A. (2007) – Mitoc Malu-
Galben : analyse archéozoologique des ateliers de
taille. In: M. Otte, V. Chirica, P. Haesaerts (dir.),
L’Aurignacien et le Gravettien de Mitoc-Malu Galben
(Moldavie Roumaine). Liège, Université de Liège
(ERAUL, 72), p.145-166.
L R.L. (2008) – Quantitative Paleozoology. Cambridge,
Cambridge University Press.
M A.E. (2004) – Measuring Biological Diversity.
Malden, Blackwell.
M L. (2009) – Geißenklösterle. Das Gravettien der
Schwäbischen Alb im europäischen Kontext. Tübingen,
Kerns Verlag, 367 p.
— (2010) – Geißenklösterle. e Swabian Gravettian in its
European Context. Quartär, 57, p.79-93.
— (2012) – Le Gravettien ancien d’Europe centrale revi-
sité : mise au point et perspectives. L’Anthropologie,
116 (5), p.609-638.
M L., B M., N P.R. (2016) – Did Prehis-
toric Foragers Behave in an Economically Irrational
Manner? Raw Material Availability and Technological
Orga nisation at the Early Gravettian Site of
WillendorfII (Austria). Quaternary International, 406,
p.84-94.
N P.R. (2004) – “Neue” alte Funde aus Willendorf II,
Schicht 3. Archäologie Österreichs, 15 (1), p.23-25.
— (2012) – e Early Upper Palaeolithic of the Middle
Danube Region. Leiden, Leiden University Press, 379 p.
N P.R., V B., T G. (2007) – KG Willendorf.
Fund berichte aus Österreich, 45, p.626-627.
N P.R., V T.B., H P., B S.,
D F., F C., F M., G M.,
H U., M C., M L., N L.,
R M., R D., Z L., T G.,
H J.-J. (2008a) – New Research on the
306 P R. N M D. B
Aurignacian of Central Europe: A First Note on the
2006 Fieldwork at Willendorf II. Quartär, 55, p.9-15.
N P.R., V B., T G. (2008b) – KG Willendorf.
Fund berichte aus Österreich, 46, p.610-611.
N P.R., V T.B., H P., T G. (2008c)
– Willendorf II. Wissenschaliche Mitteilungen aus dem
Niederösterreichischen Landesmuseum, 19, p.31-58.
N P.R., V B., T G. (2011a) – KG Willendorf
(2009). Fundberichte aus Österreich, 49, p.327-328.
— (2011b) – KG Willendorf (2010). Fundberichte aus
Österreich, 49, p.328-329.
— (2012) – KG Willendorf (2011). Fundberichte aus
Österreich, 50 (2011), p.305.
N P.R., H P., D F., F-
F C., M C., V B., G M.,
N L., T G., H J.-J. (2014) – Early
Modern Human Settlement of Europe North of the
Alps Occurred 43,500 Years Ago in a Cold Steppe-Type
Environ ment. Proceedings of the National Academy of
Sciences, 111 (40), p.14394-14399.
N P.R., L T., H P., B, M.D.,
B T., C V., N P. (2021) – e
mid Upper Palaeolithic (Gravettian) sequence of Mitoc-
Malu Galben (Romania): New eldwork between
2013 and 2016 - Preliminary results and perspectives.
Quaternary International, 587-588, p.189-209.
N P. (2009) – Le Paléolithique supérieur de Moldavie.
Essai de synthèse d’une évolution multi-culturelle. Liège,
Uni versité de Liège (ERAUL, 121), 609 p.
N P., C V., H P., V D.-M.,
B G., B T., L T., B M.D.,
N P.R. (2016) – Nouvelles recherches de terrain à
Mitoc-Malu Galben 2013-2015. In: V. Cirica, C. Ichim
(dir.), Les Aurignaciens. Leur création matérielle et
spiri tuelle, actes du colloque international de Iasi (28-
31 janvier 2016). Targoviste, Editura Cetatea de Scaun,
p.13-49.
O J., B F.G., F M., K R.,
L P., MG D., M P.R.,
O’H R.B., SG.L., S M.H.H.,
SE., W H. (2019) – Vegan: Community
Eco logy Package. R package version 2.5-4. https://
CRAN.R-project.org/package=vegan.
O M. (1981) – Le Gravettien en Europe centrale. Bruges,
De Tempel (Dissertationes Archaeologicae Gandenses,
20), 504 p.
— (1990) – Révision de la séquence du Paléolithique
supérieur de Willendorf (Autriche). Bulletin de
l’Institut royal des Sciences naturelles de Belgique,
Sciences de la Terre, 60, p.219-228.
O M., C V., H P. (dir.) (2007) –
L’Aurignacien et le Gravettien de Mitoc-Malu Galben
(Moldavie Roumaine). Liège, Université de Liège
(ERAUL, 72), 233 p.
P M., S J., H B., S S.
(2014) – Předmostí III: un site pavlovien de la porte
de Moravie (République tchèque, Europe centrale).
L’Anthro pologie, 118 (3), p.255-291.
R C T (2019) – R: A Language and Environment
for Statistical Computing. Vienna, R Foundation for
Statistical Computing.
R B.J., D H.L. (1998) – Production and Transport
of Blanks and Tools at the French Middle Paleolithic
Site of Combe-Capelle Bas. American Antiquity, 63 (1),
p.47-62.
R J.P. (1982a) – An Extension of Shapiro and Wilk’s
W Test for Normality to Large Samples. Journal of the
Royal Statistical Society: Series C (Applied Statistics), 31
(2), p.115-124.
— (1982b) – Algorithm AS 181: e W Test for Normality.
Jour nal of the Royal Statistical Society. Series C (Applied
Sta tistics), 31 (2), p.176-180.
— (1995) – Remark AS R94: A Remark on Algorithm AS
181: e W-Test for Normality. Journal of the Royal
Sta tistical Society. Series C (Applied Statistics), 44 (4),
p.547-551.
S C.E. (1948) – A Mathematical eory of Com-
mu nication. Bell System Technical Journal, 27 (3),
p.379-423.
S E.H. (1949) – Measurement of Diversity. Nature,
163 (4148), p.688.
S I.F., F P.J. (2003) – A Tribute to Claude
Shannon (1916–2001) and a Plea for more Rigorous
Use of Species Richness, Species Diversity and the
‘Shannon-Wiener’ Index. Global Ecology and Bio-
geography, 12 (3), p.177-179.
E D H-G B E M-U P 307
S M.C., K S.L. (1992) – Subsistence, Technology,
and Adaptive Variation in Middle Paleolithic Italy.
American Anthropologist, 94 (2), p.306-339.
S T.A. (2012) – Toward a Behavioral Ecology of
Lithic Technology: Cases from Paleoindian Archaeology.
Tucson, University of Arizona Press, 304 p.
S J.A., K B., J L., Š P. (1999)
– e Gravettian in Moravia: Climate, Behaviour
and Tech no logical Complexity. In: W. Roebroeks,
M. Mussi, J. Svoboda, K. Fennema (dir.), Hunters of
the Golden Age: e Mid Upper Palaeolithic of Eurasia
30,000-20,000 BP. Leiden, University of Leiden
(Analecta Praehistorica Leidensia, 31), p.197-217.
S J.A., L V., V E. (1996) – Hunters
between East and West. e Paleolithic of Moravia.
New York, Plenum Press, 386 p.
T E. (1959) – Die jungpleistozäne Wirbeltierfauna
von Willendorf in der Wachau, N.Ö. In: F. Felgenhauer
(dir.), Willendorf in der Wachau. Monographie der
Paläolith-Fundstellen I-VII. Vienna, Rohrer, p. 133-
170.
V A. (2009) – Limiting Factors on Early Modern
Human Dispersals: e Human Biogeography of Late
Pleni glacial Europe. Quaternary International, 201
(1-2), p.77-85.
W H. (2016) – ggplot2: Elegant Graphics for Data
Analysis. New York, Springer, 253 p.
W C.O. (2019) – cowplot: Streamlined Plot eme and
Plot Annotations for ‘ggplot2’. R package version 0.9.4.
https://CRAN.R-project.org/package=cowplot.