Volume 101, 1 October 2014, Pages 193–206
Wild plant use in European Neolithic subsistence economies: a formal assessment of preservation bias in archaeobotanical assemblages and the implications for understanding changes in plant diet breadth
- Open Access funded by European Research Council
- Under a Creative Commons license
Open Access
Highlights
- •
- We compare differences between charred and waterlogged samples at European Neolithic sites.
- •
- The broader range of wild plant food taxa represented in waterlogged samples is due to preservation.
- •
- Diachronic changes in edible wild taxa in the Neolithic reflect different strategies of land use.
- •
- Neolithic plant diet breadth cannot be reconstructed from waterlogged samples alone.
Abstract
In
this paper we estimate the degree to which the range and proportion of
wild plant foods are under-represented in samples of charred botanical
remains from archaeological sites. We systematically compare the
differences between central European Neolithic archaeobotanical
assemblages that have been preserved by charring compared to those
preserved by waterlogging. Charred archaeobotanical assemblages possess
on aggregate about 35% of the range of edible plants documented in
waterlogged samples from wetland settlements. We control for the
ecological availability of wetland versus terrestrial wild plant foods
on assemblage composition and diversity, and demonstrate that the
significantly broader range of wild plant food taxa represented is
primarily a function of preservation rather than subsistence practices.
We then consider whether observed fluctuations in the frequency of
edible wild taxa over time can also be attributed to preservation, and
demonstrate that it cannot; and thus conclude that there are significant
changes in plant food diets during the Neolithic that reflect different
strategies of land use and, over time, a decreasing reliance on
foraging for wild plant foods. The wild species included in our analyses
are not spatially restricted—they are common throughout central Europe.
We maintain, therefore, that our results are relevant beyond our study
area and more generally illustrate the challenges of attempting to
reconstruct the relative importance of wild plant foods—and thus plant
diet breadth—in Neolithic archaeobotanical assemblages from charred data
alone.
Keywords
- Edible wild plants;
- Preservation bias;
- Pile dwellings;
- Central European Neolithic;
- Plant diet breadth
1. Introduction: charred plant remains, missing plant foods, and Neolithic subsistence economies
The
disparity between the quantity and range of remains preserved in
waterlogged and charred form has long been recognised. This is apparent
with regard to both crops and wild taxa and, for example, records of
waterlogged specimens can outnumber those of charred specimens by
hundreds, or even thousands (Brombacher and Jacomet, 1997, Tables 36 and 37; Heer, 1865 and Jacomet, 2013, 501; Jacomet et al., 1989, Tables 32–34, Figs. 49 and 50; Jacomet et al., 2004,
Figs. 84 and 85). Despite this, reconstructions of European Neolithic
plant-based diets have unavoidably tended to use the more durable
macrofossils, mainly charred seeds and grains of domestic crop species,
to make inferences about food choices that emphasise the contribution of
cultivars to total diet. Bogucki (2000,
204), for example, states: “The earliest farming communities in
north-central Europe grew a suite of crops, including emmer and einkorn
wheat, barley, peas, flax and poppy… A recurring set of weeds of
cultivation occurs among the carbonized seed samples, but otherwise no
wild plants figure prominently in the subsistence of the early farmers.”
(for further examples see papers in: Colledge and Conolly, 2007, Douglas Price, 2000, Fairbairn, 2000, Milles et al., 1989 and Van Zeist et al., 1991).
However charred remains represent only a small and biased sample of the
edible plant taxa used by Neolithic societies. Our ability to establish
the composition and overall contribution of plant-based diets in
subsistence is thus biased and inherently limited because of this
largely unavoidable reliance on charred macrofossils. Hillman refers to
the undetectable component of the diet as the ‘missing foods’ (Hillman, 1989),
and we here present our attempt to identify the extent to which our
reconstructions of food choices are biased on the basis of a comparative
analysis of waterlogged versus charred plant assemblages from central
European Neolithic sites.
1.1. Formation and interpretation of archaeobotanical samples
In
comparison to charring, preservation by waterlogging in anoxic or
anaerobic conditions results in less taphonomic bias and the composition
of waterlogged samples is likely to resemble more closely the original
suites of plants utilised (Jacomet, 2013, 500; Willerding, 1971,
181). There is little or no discrimination (e.g., on the basis of
robustness of plant components) as to what is preserved: soft fruits,
underground storage organs and leafy vegetables, together with more
fragile plant parts, such as pericarp, leaves, flowers, catkins, etc.,
that rarely survive after exposure to heat remain intact and, by
comparison, overall losses are relatively small.
Pétrequin's
ethnographic accounts of the deposition and accumulation of debris from
daily activities at the pile-dwelling villages of the Toffinu on Lake
Nokoué in the Republic of Benin are pertinent to the present study (Pétrequin, 1986).
The Toffinu (“men of the water”) were previously dryland farmers who
subsequently adapted to living on marshes and lakes in pile-dwellings.
Pétrequin notes that food waste (including nuts, seeds, grains, shells
and fish bones) was thrown into the water through trap doors in floors
in the kitchen areas (Pétrequin, 1986,
65). Likewise, human excrement was discharged via openings in houses
directly onto the waterlogged sediments below. Rubbish accumulated into
dumps that formed artificial islands and use was made of these as small
livestock pens. Waste from several families would often form larger
islands that provided shelter for larger animals such as cattle. There
was a process of sorting on/in the sediments according to density or
size of the organic elements ejected from the houses. Pétrequin
describes patterns of deposition in layered lenses of a variety of
anthropogenic debris at different depths of water such that heavier
items (logs, branches) settled further out into the lake and lighter
ones closer to the shore (Pétrequin, 1986:
Fig. 14). The cycle of deposition of organic materials, accumulation
and incorporation in waterlogged sediments continued after houses were
abandoned and had collapsed into the water, contents (e.g., tools that
were hung from walls) often settled in exactly in the same place as they
had done when the houses were occupied.
Changes
in the chemical balance of the water or waterlogged sediments (e.g.,
addition of chemical contaminants, changes in the availability of oxygen
or increased acidity or alkalinity) can, however, initiate the
decomposition process and also create conditions in which bacteria that
cause organic decay are activated (Brinkkemper, 2006, Caple and Dungworth, 1998 and Holden et al., 2006).
Decomposition of plant materials may be accelerated in near surface
sediments, which are more prone to periodic drying-out and higher rates
of oxygen diffusion than those at greater depths, however there appears
to be no consensus as to whether or not this is the case for all sites
(e.g., urban versus rural settlements; Caple and Dungworth, 1998, Section 3.7; Carrott et al., 1996, 7; Jones et al., 2007, 83; Kenward and Hall, 2000 and Willerding, 1991,
31). Moreover, waterlogged deposits of plant remains are relatively
rare across Neolithic Europe and are restricted to the Alpine lakes and
very few other wetland areas in northern Europe (e.g., northern Germany:
Kirleis et al., 2012 and Kroll, 2007; Netherlands: Kubiak-Martens, 2006, Kubiak-Martens, 2012, Out, 2009, Raemaekers et al., 1997 and Van Haaster, 2001).
Neolithic wells are as rare in central Europe but several structures
have been found with stratified waterlogged deposits (e.g., below the
water table) and from which diverse assemblages of plant remains have
been recovered (Zerl and Herbig, 2012).
Plant
materials used as fuels are a common source of charred remains, as are
stored products that are destroyed in accidental or deliberate fires
(e.g., to dispose of infested crops, or as a result of acts of violence;
Van der Veen, 2007,
979). In addition, charred remains typically derive from the
unintentional burning of plant foods and plant materials during
preparation, processing or cooking (Hillman, 1981,
139–140). Modern charring experiments have demonstrated that the
ability of plants to survive is dependent on their morphology (i.e.,
both physical and chemical) and on the construction, temperature and
duration of the fires (Boardman and Jones, 1990, Guarino and Sciarrillo, 2004, Gustafsson, 2000, Wilson, 1984, Wright, 2003 and Märkle and Rösch, 2008).
Dry seeds with low densities combust more readily when heated and are
more likely to survive intact than those that are fresh and have higher
densities. Equally likely to survive are specimens with low oil content (Wilson, 1984, 204–205; Wright, 2003,
578). The experiments also demonstrate that plant materials burned in a
reducing atmosphere (e.g., buried in the deposits underlying the fire
or completely embedded within its fabric) can withstand higher
temperatures for longer than under oxidising conditions and that
exposure to intense heat over long periods results in the greatest
losses (Boardman and Jones, 1990, Fig. 1; Wilson, 1984, Table 6; Wright, 2003, 578). Cereal chaff was shown to be less durable than grains at high temperatures (Boardman and Jones, 1990,
4–6 and Fig. 1); other less robust plant parts, such as soft fruits,
leaves, roots/rhizomes/bulbs (etc.), are rarely preserved in charred
assemblages (Willerding, 1971 and Willerding, 1991).
The
seeds of some species are able to withstand considerable additional
heat after charring and prior to total destruction whereas others have a
much narrower range of survival and are thus more fragile in charred
form. In oxidising conditions, for example, cereals tolerate greater
increases in temperature than oil rich seeds such as Linum usitatissimum (domestic flax) and Papaver somniferum (opium poppy) before being reduced to ash ( Boardman and Jones, 1990, 4–5; Märkle and Rösch, 2008,
S260–261). The fragility of the plant components is highlighted by the
fact that proportional losses after burning have been shown to be great:
in experimental firing of hearths and houses between 60 and 80% of
cereal grains failed to survive and there were similar, if not greater
losses for seeds of wild species ( Guarino and Sciarrillo, 2004, Figs. 3, 6 and 11; Gustafsson, 2000,
Figs. 3 and 6). Archaeological samples after burning are therefore
likely to comprise species with more resilient seeds and the probability
is low that their overall composition bears direct relationship to the
original taxa proportionality or diversity. Incomplete burning resulting
in charring is a means by which only a subset of the plants originally
used at a settlement survives and partial destruction is a more apt
description of the process than preservation (cf. Wilson, 1984, 201).
In
spite of these taphonomic limitations the analytical scope of charred
assemblages is considerable. Charred remains dominated by crops (e.g.,
cereals and pulses) and wild plant species (e.g., mostly those common to
cultivated fields) are ubiquitous on Neolithic and later sites across
Eurasia (Hillman, 1981, Jones, 1984, Jones, 1988 and Knörzer, 1971, 90–91; Van der Veen, 1992,
77). Their great interpretative strength hinges on the overall
representational consistency of crops and wild species, thus
facilitating spatial and temporal comparisons between sites or groups of
sites (for example, see papers in Colledge and Conolly, 2007).
The frequency with which the charred assemblages occur has enabled the
development of sophisticated models concerning all aspects of arable
farming systems: from the apportioning of responsibilities for
production (at inter- and intra-site levels), the type of cultivation
systems implemented, the organisation and scheduling of labour, the pre-
and post-harvest tending of fields and crops, and the storage,
processing and preparation of crops prior to consumption (whether by
humans or animals). The basis for the interpretative models is the
quantitative assessment of inter-assemblage compositional variability in
the proportions of grains, chaff and seeds of wild plants.
The
similarity was noted at early Neolithic Bulgarian sites, in terms of
the ratios of grains to chaff and grains to weed seeds, between
archaeobotanical samples and the various stages in the sequence of crop
processing leading up to and including storage and cooking (Dennell, 1972 and Dennell, 1974).
The significance of being able to identify the type of processing
activities that took place was recognised, as was the necessity to take
into account the effect of taphonomic filtering (e.g., as crops are
cleaned and chaff and weed contaminants are removed by winnowing,
sieving, etc.) prior to reconstructing subsistence economies (Dennell, 1974, 283). Hillman's seminal works in the 1970s, 1980s (Hillman, 1973, Hillman, 1981 and Hillman, 1984), based on his ethnographic studies at agricultural settlements in the Aşvan region (Elâziğ
province) in Turkey, strengthened this comparative approach by
establishing a direct association between the modern crop processing
discard recorded at the village farms and the archaeological samples.
More specifically the uniqueness in composition of crop processing
products and by-products enabled accurate categorisation of the
archaeobotanical assemblages and the cultivation methods from which they
derived, for example, including details of tillage and harvesting
techniques, sowing times and post-harvesting processing stages prior to
the final preparation of the crops for consumption (Hillman, 1984, Jones, 1984 and Jones et al., 2005).
The potential to classify ancient farming systems in terms of arable
versus pastoralist economies, or producer versus consumer settlements,
and to distinguish between intensive and extensive forms of agricultural
production was realised by the use of ecological methods to assess the
composition of weed species in charred assemblages (Bogaard, 2004, Jones, 1988 and Van der Veen, 1992; see also Stevens, 2003 and Van der Veen and Jones, 2006).
In areas where woodlands are sparse and fires are of necessity fuelled
by animal dung it has also been possible on the basis of charred seeds
and cereal chaff preserved within matrix of the dung to gain insights
into aspects of animal husbandry (e.g., livestock diets, mobility and
seasonality of herding) and to what extent arable and pastoral economies
are integrated to ensure sustainable systems of production for both
crops and livestock (Bottema, 1984, Charles, 1998, Miller, 1984, Miller, 1999, Miller and Smart, 1984 and Wallace and Charles, 2013).
1.2. Problem definition
We
seek to quantify and to understand the source of variation in the range
of wild plant use at Neolithic sites in central Europe. In particular,
we wish to establish the degree to which samples of charred remains are
correlated with low wild plant taxa diversity and samples of waterlogged
remains with high plant taxa diversity and whether or not any
correlations can be explained either by taphonomy or as a function of
different types of foraging opportunities. To answer this question we
compare taxonomic similarities and differences at a representative
sample of Neolithic sites at which preservation of plant materials by
both charring and waterlogging is recorded. We attempt to control for
variation in foraging opportunity by selecting sites that are all
adjacent to water or wetlands (e.g., settlements on or near lakes,
watercourses and marshes) so we can a priori anticipate that they share a
similar range of available plant species.
