-
1
Bone Chemistry Analysis:
The Theoretical Basis
Introduction
Although many tissues of the human body can be used to determine
diet and nutritional status (Underwood
1977), bone is often times the only tissue recovered from an
archaeological context. Due to its availability
the majority of archaeological dietary studies have used bone as
the material for analysis (for an overview
see Price 1989; Price et al. 1985b; Schoeninger and Moore 1992;
Schwarcz and Schoeninger 1991). Bone is
a complex cellular tissue that is composed of three major
components: an organic matrix (mostly collagen),
an inorganic mineral fraction, and water. Both the organic and
inorganic components are used in conducting
bone chemistry analysis; the inorganic mineral fraction is used
for elemental analysis, and the organic
portion (collagen) is necessary for isotopic research.
The following chapter will outline the utility of isotopic and
elemental analyses in reconstructing diet, and
examine how these analytical procedures can be used effectively
in archaeological research. Stable carbon
and nitrogen isotopes have been used to identify differences in
the consumption of meat, specific terrestrial
plant foods (such as maize), marine plants and mammals. Trace
element analysis can distinguish relative
differences in the consumption of meat, plant foods, and marine
resources. The post-mortum alteration of
bone (known as diagenesis) will also be discussed in light of
its ability to alter the biological quantities of
isotopes and trace elements, thereby inhibiting dietary
reconstruction. While diagenesis is pervasive in all
archaeological bone, it can be controlled through several new
techniques that are discussed at the end of this
chapter.
Stable Isotope Analysis
Carbon Isotopes
The organic matrix of bone is composed mainly of collagen
(approximately 90%), followed in abundance
by noncollagenous proteins, lipids and carbohydrates (Boskey and
Posner 1984). Present within the bone
collagen are stable isotopes of carbon and nitrogen, that are
frequently used to reconstruct ancient diet (for
reviews see Schoeninger and Moore 1992; Sihlen and Kavanaugh
1982). While the amount of carbon and
nitrogen in the body is under strict homeostatic control, the
ratios of their stable isotopes (13C/12C and 15
in bone collagen vary according to the ratios found in food
items and in the surrounding environment.
The differences in the isotopic ratios of carbon and nitrogen
are due primarily to environmental differences
and biochemical or physiological reactions of organisms
(Schwarcz and Schoenmger 1991). During these
reactions, one isotope is discriminated against in favor of the
other, resulting in a change in the ratio of the
carbon and nitrogen isotopes involved. These ratios are
expressed relative to a standard and in par per
million (%o) using the following formulas:
The standard used for carbon is PeeDee Belemnite Carbonate (PBD)
a marine fossil limestone from South
Carolina (Craig 1957). The standard used for nitrogen is
at.mospheric N (AIR), since its isotopic ratio is
constant worldwide (Mariotti 1980).
All terrestrial plants rely on atmospheric CO2 as their major
carbon source but differentially fix CO2
according to one of three photosynthetic pathways: C3 C4 and CAM
(crassulacean acid metabolism). The C3
and C4 pathways are so named because during the first stage of
photosynthesis C3 plants produce a molecule
containing three carbon atoms, while C4 plants produce a
molecule with four carbon atoms (OLeary 1981). C3 plants include
wheat and rice, all root crops, vegetables, legumes, nuts, most
fruits, and a rnajority of
temperate zone grasses. C4 plants are represented by the
tropical grasses. C4 plants are represented by the
tropical grasses, such as millet, sorghm maize, some amaranths,
and some chenopods. Tropical succulents,
-
2
such as pineapple and various consumers of the next trophic
leveis, such as carnivores or omnivores will
show an additional isotopic fractionation of approximately 1 %o
from their chosen diets (Bender et al. 1981;
DeNiro and Epstein 1978b; Schoeninger 1985). Taking into
consideration the above mentioned collagen enrichment factor and
additional 1 %o trophic level fractionation, Chisholm (1989)
provided a table of expected values for several different dietary
regimes (Table 3.1). From Chisholms expectations, it would be
difficult to distinguish individuals consuming large amounts of
maize (-7.5% from those heavily dependent
on meat from C4-eating herbivores (-6.5%o).
Table 3.1. Anticipated a13C Values for Consumers and their
Specific Diets (from Chisholm 1989:Table 2.1)
Ave. Dietary C (%) Consumer a13C (%)
C3 plants only -26.5 -21.5
Meat from herbivores on C diets -25.5 -20.5
C plants only -12.5 -7.5
Meat from herbivores on C diets -11.5 -6.5
Marine plankton only -19.5 -14.5
Meat from marine herbivores -18.5 -13.5
Meat from marine carnivores -17.5 -12.5
It appears then, that differential discrimination of carbon
isotopes by living organisms allows for the
reconstruction of their dietary components. One of the first
scholars to realize the potential of differential
fractionation by plant species as a method of reconstructing
cacti are members of the CAM plants, and fix
atmospheric CO2 by either a C3 or C4 pathway depending on their
environmental conditions. Some CAM
plants, such as Opuntia, rely almost entirely on the C4
pathway.
The biochemical differences in these pathways result in isotopic
fractionation and 13C values that are
distinctive according to plant type. For example, C3 plants have
13C values ranging between -20 and -34%o,
with an average of -26.5%o. Modern C4 species range between -9%o
and -16%o, and average near 12%o.
CAM plants that thrive m sunny, arid microhabitats fix CO2 in
the same manner as C4 plants, and thus, have
a values similar to C4 species (OLeary 1481,1988; Smith and
Epstein 1971). Although sorne environmental factors such as water
availability, temperature, light intensity and available soil
nutrients can effect the a13C
values of C3 plants, the degree of fractionation is so distinct
from C4 plants that their is no overlap in their
a13C values (Farquhar et al. 1982; Tieszen 1991).
When an animal consumes plant foods, the carbon isotopes in the
plant are further fractionated during
metabolism, causing a change in the animals a13C value (Bender
et al. 1971; DeNiro and Epstein 1978a, b; van der Merwe and Vogel
1978). This change, known as the collagen enrichment factor,
produces a 13C bone collagen value approximately 5%o more positive
than the plant food originally consumed (Chisholm
1989; Chisholm et al. 1982,1983; Krueger and Sullivan 1984). For
example, if an herbivore consumed only
C4 grasses, its bone collagen 13C value would be near -7.5%
(-12.5 +5%o). It has also been reported that
49
ancient diet was Robert Hall (1967a, 1967b, 1967c). In an
attempt to explain anomalous C dates taken from
grass and com samples, Hall found that different plant species
had characteristic 13C/12C ratios, that could be
-
3
used to distinguish browsing animals (i.e. bison) from grazers
(i.e. deer) (1967a: 5). Hall concluded his
paper with a great deal of insight, when he suggested that;
Presumably the diet of humans would be affected in sorne similar
way. This opens up an obvious lime of investigation because of the
reliance of sorne people on corn, others on bison flesh, others on
acrns others
on marine sources and others on various combinations of these...
(Hall 1967a: 5-6).
Seventeen years after Halls initial work, carbon isotope ratios
have become so valuable to the reconstruction of ancient diets,
that they are used in the majority of bone chernistry research. In
addition to
carbon, the isotopes of nitrogenalso found in bone collagenhave
been used with varying degrees of success to reconstruct past diet.
Foliowing is a discussion of nitrogens value as a paleodietary
indicator.
Nitrogen Isotopes
Nitrogens two stable isotopes, 15N and 14N are found almost
exclusively either dissolved in the oceans or bound as N2 in the
atmosphere (Marjotfi 1983). The ratio of these nitrogen isotopes
can be used to
distinguish organisms that are N2-fixing from those that
utilize
other forms of nitrogen, such as ammonia, soil nitrates, or
animal urea (Ambrose 1993; Schoeninger &
Moore 1992). The biochemical reaction that take place in
N2-fixing organisms produce 15N values very
similar to atmospheric N2 which approximates zero. Due to the
greater amount of 15N relative to 14N soil
nitrates, ammonia and animal urea, plants that rely on these
nitrogen sources have more positive 15N values
than N2 fixing plants. Thus, N2-fixing plants such as legumes,
have lower 15N values than other terrestrial
plants, such as grasses (Delwiche & Steyn 1970; Wada et al.
1975). Many factors effect the degree of
nitrogen isotope fractionation in plants, including c1imate
(Heaton 1987; Shearer & Kohl 1986), fertilizer
use (Aufderheide et al. 1988), soil type (Shearer et aL 19&
and even altitude (Ambrose 1993). For this
reason, terrestrial plants Show (a wide range of 15N values,
although many species have isotopic values
close to zero (Wada et al. 1975).
Just as there is a collagen enrichment factor for carbon
isotopes, the same appears to hoid true for nitrogen isotopes. The
amount of enrichment between diet and bone, however, is somewhat
unclear. Studies done on
pigs and mice in a laboratory setting (DeNiro & Epstem 1981;
Hare et al. 1991) and on plants and animais
in terrestrial and marine ecosystems (Minagawa & Wada 1984;
Vogel et al. 1990; Wada 1980), suggest that
bone collagen is approximately 3-4% more enriched in 15 than
diet Research done on free-ranging
herbivores, however, illustrate that bone collagen enrichment
may vary considerably on account of
environmental and physiological factors (Ambrose 1986; 1991;
Ambrose and DeNiro 1986; Heaton et al.
1986; Schoeninger 1989; Sealy et al. 1987).
There also seems to be a step-wise enrichment of 15 withm a
single trophic system (McConnaughey &
McRoy 1979). The 5 values increase by approximately 3%o as one
moves up the food cham. This step-wise
enrichment has been demonstrated for both marine and terrestrial
vertebrates and invertebrates (Schoeninger
1985, 1989; Schoenmger & DeNiro 1984; Wada 1980).
If marine food items have 15 values that are distinct from
terrestrial food items, and terrestrial N2-fixing
plants can be distinguished from those that do not fix
atmospheric nitrogen, then it should be possible to
determine the presence of marine foods and legumes (N2-fixing)
in the human diet. Many studies have
identifled populations dependent primarily on marine foods from
those dependent primarily on terrestrial
foods (Schoeninger et al. 1983, 1990; Medaglia et aL 1990;
Walker and DeNiro 1986).
Studies attempting to demonstrate legume consumption, however,
have been less successful (Minagawa and
Akazawa 1991; Schwarcz et al. 1985; White and Schwarcz 1989;
Spielinan et al. 1990). Since legumes
usually have lower values, it is assumed that a heavy reliance
on legumes produces lower than normal 3 bone collagen values.
Unfortunately, to date there are no laboratory experiments that
substantjate this claim.
-
4
In addition, Spielman and co-workers (1990) analyzed
archaeological beans and found that the 15 values
were much more positive than expected.
These elevated 15 legume values may be a consequence of the
re-working of agricultural fields. When a
fleid is prepared for cultivation soil nitrates become more
readily available and the 15N value of the soil
increases. With each successive use of the agricultural field,
the soil 15N value continues to rise (Mariotti et
al. 1983). As these soil nitrates become available, legumes stop
fixing atmospheric nitrogen and begin to
utilze this new nitrogen source, producing 15N values that are
more postive than atmospheric nitrogen
(15Na near O %o) (Shearer et al. 1983). Thus, the elevated 15N
values present in prehistoric bean samples
may represent plants grown in cultivated fields and utilizing
soil nitrates rather than fixing atmospheric
nitrogen.
The 3%o enrichment of 15 by trophic level raised hopes among
scholars that nitrogen isotopes could be
used to identify the consumption of animal protein by ancient
human populations (Schoeninger 1985).
Unfortunately, since humans are neither exclusively carnivorous
nor herbivorous, the 3%o 15N enrichment
factor is not detectable. Instead, most humans groups consume a
percentage of animal protein in their diet,
ranging from approximately 10% to 25% in human societies world
wide (Schoeninger and Moore 1992). As
Schoeninger and Moore (1992) illustrate, there is only a O.5%
difference in the 15N values of those
consuming 10% animal protein and those consuming 25%.
Consequently, the very small difference in 15
values limits nitrogens use in determining meat consumption
between human populations.
It appears then, that nitrogen isotopes are of rather limited
utility. While nitrogen can be used to distinguish
between marine and terrestrial components, it has been less
effective as an indicator of meat and/or legume
consumption. Carbon isotopes, on the other hand, have been used
more successfully to reconstruct ancient
dietary components. A new approach suggests that carbon ratios
can even be more reliable in reconstructing
diet when taken from both bone collagen and bone apatite. The
foliowing section outlines this new
technique.
Collagen versus Apatite as a Paleodietary Indicator
While many of the studies discussed aboye rely on a predictable
relationship between collagen and the
isotopic composition of diet to differentiate between social
groups, collagen may not accurately reflect the
sum importance of the lipid, carbohydrate and protein fraction
in the diet (Ambrose 1993; Chisholm 1989;
Chisholm et aL 1982; Krueger and Sullivan 1984). Carbon atoms
used to synthesize collagen may come
from different dietary fractions according to the amount of
protein consumed. According to a model
proposed by Krueger and Sullivan (1984), when protem intake is
low, protein carbon will be deposited in
bone coilagen, since it is used primarily for growth and
maintenance of bodily tissues (Krueger and Sullivan
1984). Lipids and carbohydrates, on the other hand, are used
primarily for energy metabolism (DeNiro and
Epstein 1978a). When lipids and carbohydrates undergo cellular
metabolism, the associatecj carbon atoms
are synthesized into blood bicarbonates, after which they are
incorporated into the apatite fraction of bone
(DeNiro & Epstein 1978a). Consequently, since the protein
fraction appears to contribute carbon atoms to
collagen, while carbohydrate and lipid carbon settles in bone
mineral, it has been suggested that only the
protein fraction of the diet will be discernible in collagen
carbon isotope ratios (Chisholm et al. 1982).
Krueger and Sullivan (1984) further hypothesized that if
different dietary fractions were responsible for
collagen and apatite synthesis, than the ratio between apatite
and collagen a 13C values (13CA-C would be
smaller in carnivores than in herbivores, due to a differential
emphasis on lipids, carbohydrates and protein
in the diet. Herbivores acquire protein primarily from plant
protein, and energy from plant carbohydrates,
while carnivores utilize animal flesh for protein, and lipids
and animal protein for energy metabolism. Since
lipids are low in 13C relative to carbohydrates and protein
(DeNiro & Epstein 1978a; Jacobson et al 1972;
Winlder & Schmidt 1980), and carnivores have a greater
reliance on 13C depleted lipids for energy
-
5
metabolism than herbivores, then carnivores will have lower
13CA-C values. Thus, it was hypothesized that
while carnivores had the lowest 13CA-C L values, and herbivores
the highest, human populations should fall
between this range depending on their meat intake. Throughout
their research, Krueger and Sullivan (1984)
implied that collagen 13C values reflect primarily protein
intake, and apatite 13C values reflect the energy
portion of the diet.
New data generated by Ambrose and Norr (in prep., as cited in
Ambrose 1993), suggest that this division
between protein and collagen, and lipids carbohydrates and
apatite, may be more complex than reported. In
a recent controlled, laboratory study, Ambrose and Norr found
that collagen carbon reflect protein intake
only when adequate protein is available in the diet for collagen
synthesis. When the diet is very low in
protein, however, carbon atoms from both lipids and
carbohydrates form the major component of bone
collagen. In other words, the origm of collagen carbon will
depend directly on the quantity of protem
consumed by an individual. Consequently, while collagen 13C
values have been used to reconstruct levels of
C3 and C4 (maize) plants in the diet (for a review see
Schoeninger and Moore 1992), this is only an accurate
predictor when low-protein diets are consumed. When enough meat
is included in the diet to supply protein
for collagen formation, the collagen 13C will reflect the animal
flesh consumednot the plant material. As Ambrose explains, ... in
diets with levels of animal protein sufficient for collagen
synthesis, carbohydrate and lipid carbon may be severely
underrepresented (Ambrose 1993:105).
In addition, contrary to the view that bone apatite 13C values
represent only the lipid and carbohydrate
fraction of the diet (DeNiro & Sullivan 1978a; Krueger and
Sullivan 1984), Ambrose and Norr found that,
collagen reflects mostly the protein 13C value, while the
apatite reflects the whole diet 13C value, not simply the energy
component (Ambrose 1993:109). Thus, when the collagen 13C value is
compared to the whole diet, apatite 13C value, both trophic levels
and dietary fractions can be distinguished (Ericson et al.
1989; Ezzo 1993; Lee Thorp et al. 1989).
A comparison of collagen and apatite 13C values may further aid
in identifying status-based dietary
differences of ancient populations. While collagen carbon has
been the primary source of a values in most
research, it reflects only specific fractions of the diet.
Further, these fractions (lipids, carbohydrates,
proteins) will be differentially reflected in collagen depending
on the level of protein intake. For example, if
two sub-populations both consume 70% of their daily calones from
maize (a C4 plant), yet only one group
has su dietary intake of a C3 protem source, the 13CA-C values
will not be equivalent between the groups. It is
likely that only the protein-deficient group will have collagen
with a C4 signal, since the collagen carbon
was derived primarily from the maize carbohydrates and lipids,
thereby reflecting a C signal. In the collagen
of the C3 protein eaters, carbon from maize will be
underrepresented, since their collagen carbon comes
from the C3 animal source. If the apatite values reflect the
entire diet, not just the protein source like
collagen, thai a comparison of the collagen and apatite 13C
values would allow for a more thorough
comparison of the differential reliance of dietary fractions
throughout the population.
One of the potential problems with bone apatite is its
susceptibility to post-mortem alterations (diagenesis)
(Schoerunger & DeNiro 1982; Land e al. 1980; for reviews see
Lee Thorp 1989; Klepinger 1984).
Biological apatite contains carbon in the form of carbonate ions
(Lowenstam & Weiner 1989), which can be
found in two primarv locations: (1) within the crystal lattice,
as a substitute for Sr, Ba, Pb, hydroxide or
phosphate ions (Baud & Very 1975; Le Geros et al. 1967), and
(2) absorbed in the hydration shell or on the
crystal surface (Baud & Very 1975; Eanes & Posner 1970).
Due to the propensity of ionic exchange at the
crystal surface, carbonate jons exposed to the surface are
readily exchanged with other ions in bodily fluids
(Newesly 1989; Piepenbrink 1989). In a post-mortem enviromnent,
the surface (absorbed) carbonate in
archaeological bone has a tendency to be replaced with ions from
the surrounding soil or groundwater.
A number of scholars have argued that a chemical pretreatment of
bone apatite with an acid, such as acetic
acid, will remove most diagenetic effects, leaving only the
original 13c/12c ratio (Lee Thorp 1989; Lee
Thorp et a!. 1989; Ericson et al. 1981; Haynes 1968; Krueger and
Sullivan 1981). It appears that an acid
-
6
pretreatment should substantially reduce diagenetic effects,
since researchers who analyze for trace
elements, and thereby rely exclusively on bone mineral have been
using similar techniques with great
success (Price et al. 1991; Sillen & Kavanaugh 1982).
While the potential for using apatite for carbon isotope ratios
appears promising, questions still remain
concerning dietary fractions and their influence in the
formation of both bone collagen and apatite. Until
more controlled, laboratory feeding experiments are conducted,
carbon isotope ratios of bone apatite should
be interpreted with caution. The above discussion, however,
assumes that the bone collagen used for
analysis is clean and well-preserved. What happens when
archaeological bone is buried for thousands of
years? How does post-mortem alteration (diagenesis) affect bone
collagen, and in turn, carbon and nitrogen
isotc ratios? The following section will explore some of the
problems associated with diagenesis and
isotopic analysis.
Diagenesis iii Isotopic Analysis
Scholars eager to apply isotopic analysis to the reconstruction
of ancient diets have only very recently
considered the role of diagenesis. White & Hannus (1983)
were among the first scholars to recognize
diagenetic processes in the organc fraction They suggested that
the chemical weathering of collagen
(initiated by micro-organisms) was the first stage in the
decomposition of bone tissue. It was unclear,
however, the minerals released during collagen diagenesis
existed via groundwater or remained and
recrystallized within the mineral fraction. Thus, the movement
of minerais m the organic fraction could
render the minera! fraction vulnerable to diagenesis.
Several scholars have examined diagenetic effects on both the
organi and inorganic bone fractions
(Schoeninger & DeNiro 1982; Nelson et al. 1986). Nelson and
co-workers analyzed a series of terrestrial
and marine mammals for elemental strontium, strontium isotopes,
carbon ratios from collagen, and carbon
ratios from apatite. They found that diagenesis was more
pronounced m the mineral than in the orgamc
fraction; marir and terrestrial samples could not be
differentiated using the carbon isotol ratios in apatites,
however, carbon ratios from collagen were successful in
separating the animal types in both modern and
archaeological assemblages.
In an attempt to identify the purity of collagen, and hence, the
extent of diagenesis, several scholars have
analyzed the amino acid composition and the C/N ratio of ollagen
(DeNiro 1985; DeNiro & Weiner 1988;
Hare 1980; Hare & Estep 1983; Hare et al. 1991; Kennedy
1988; Schoeninger et al.1989; Tuross et al.
1988). Unlike other animal tissues, collagen is unique in
possessing the amino acid hydroxyproline and
glycerin in excess of 30% (Hare 1980). Since each amino acid has
a specific stable isotope ratio, the loss of
one amino acid can greatly affect the isotopiv ratio of collagen
as a whole (Hare & Estep 1983). According
to Hare et al. (1991), isotopic analysis should be conducted on
a single collagen amino acid rather than
collagen as a whole. They argued that single amino acid could be
easily tracked through the food chain,
since the isotopic composition of a molecule records its
biochemical history in the food chain (1991:286). Results from
their analyses show that carbon isotope rafias taken from
individual amino acids were almost
identical to those fornid in the diet.
Most collagen samples used for isotopic analysis have an atomic
C/N ratio that normally ranges between 2.9
to 3.6 (DeNiro 1985). Although some of this variafion may be due
to experimental error in measuring
calcium and nitrogen concentrations, the higher values may also
indicate collagen contamjnation. According
to Kennedy (1988), C/N ratios higher than 3.4 may be the result
of contamination by humic acid, lipids,
carbonates or other carbon-rich rnaterjals. While these
contaminants may cause a shift in 13C values away
from the original biological signal, Ambrose (1993) argues that
samples with C/N ratios aboye 3.4, shoul not be rejected unless the
isotopic composition of the contaminant differs greatly from that
of the tissue
analyzed (p.75)
-
7
Although the mineral fraction of bone is more susceptible to
diagenetic forces than the organic fraction,
collagen preservation can vary greatly within a particular site.
Tests assessing the integrity and purity of
colla should be conducted before the resulting isotopic ratios
are used to reconstruct paleodiets. Amino acid
assays and C/N ratio of collagen have been successful in
identifying diagenesis in the organic bone fraction.
all methods must be used on every sample; Schoeninger (1989)
suggest that bone samples with less than
25% organic content by weight should use atomic C/N ratio to
assess collagen integrity, while those with
less than 10% organic by weight should utilize an amino acid
assay.
Trace Elements
Introduction
Unlike stable isotope analysis, trace element analysis makes use
of the inorganic portion of bone. This
inorganic, mineral fraction is compose primarily of calcium and
phosphate, but also includes a number of
minor and trace elements that are incorporated into the lattice
structure of the crystals during bone
manufacture (Lowenstam and Werner 1989). WhiJ the amount of
calcium and phosphorus are tightly
regulated in the body quantities of many trace elements vary
according to physiological and biochemical
processes. Interest in the movement of trace elements through
the human body and biosphere was largely a
consequence of research conducted on the by-products of nuclear
testing during the 1950s. The radioactive isotope Strontium 90 was
considered harmful to human health, and as such, its concentrations
were
carefully recorded throughout the environment, foodwebs, and
human and animal tissues (Alexander &
Nusbaum 1959; Alexander et al. 1956, Bowen & Dymond 1955;
Comar 1963; Comar et al. 1955; Elias et al.
1982; Harrison et aL 1955; Odum 1951, 1957; Steadman et al.
1958; Taylor et al. 1962; Thurber et aL 1958;
Turkian & Kulp 1956; Wasserman & Comar 1956).
These initial studies demonstrated that strontium concentrations
varied according to physiological and
biochemical processes such that plant species were identifiable
by strontium content. Toots and Voorhies
(1965) were the first to apply these data to distinguish grazmg
fossil herbivores from browsers by measuring
bone strontium leveis. However, it wasnt until the 1970s that
anthropologist began to recognize the potential of trace element
analysis to reconstruct paleodiets (Boaz & Hampel 1978; Browns
1973; Gilbert
1975; Kavanaugh 1979; Lambert et al. 1979; Schoeninger 1979a, b;
Szpunar 1977; Szpunar et al. 1978).
Trace element analysis heid great promise, because rather than
rely on artifacts or floral and faunal remains
to indirectly estimate diet, this analysis was capable of
measurrng the skeletal material directly.
As interest in trace element analysis continued to grow, it
gained wide spread acceptance for reconstructing
ancient subsistence strategies (Schoeninger 1981; Sillen 1981a,
b; Katzenburg 1984; Edward et al. 1984;
Bisel 1980; Blakely & Beck 1981; Brown & Blakely 1985;
Connor & Slaughter 1984; Geidel 1981, 1982;
Gilbert 1985; Harch & Geidel 1983; Katzenburg & Schwarcz
1984; Lambert et al. 1982, 1983; Price 1985,
1986; Frice & Kavanaugh 1982; Price et al. 1985a, b,
Schoeninger & Peebles 198 While many trace
elements were used to reconstruct diet, most of these studies
focused on the unportance of strontium, due to
its theoretical ability to distinguish between meat and
vegetable components in the diet This theoretical
foundation is presented below.
Theoretical Basis
Trace element analysis rests on an understanding of bone
microstructure and physiology. Mature bone
mineral is composed primarily of a crystalline form of calcium
phosphate, known as hydroxyapatite. With
an unit celi formula of Caio (PO4)6(OH)2 calcium plays a major
role in the formation of hydroxyapatite
(Neuman & Neuman 1958; Sillen 1989; Baron 1990). Bone
hydroxyapatite can fluctuate from this predicted
structural formula, however, during an ionic exchange between
body fluids and the bone minera! (Neuman
& Neuma 1958; Pate & Hutton 1988; Sillen 1989; Pate et
al. 1991).
Due to the large, highly-charged surface area of hydroxyapatite
crystals, various ions in body fluids (such as
strontium, barium and lead) are attracted to the crystal and
easily incorporated by displacing the calcium
-
8
component (Neuman and Neuman 1980; Johnsson 1986; Molleson 1990;
Pate et al. 1991; McLean & Urist
1955; Ortner & Putschar 1985; Underwood 1977). As a result,
rather than being a pure, insoluble crystalline
form, trace amounts of strontium, barium and lead are commoniy
found within the lattice structure of the
hydroxyapatite crystal.
During mammalian metabolism, a proportionateiy greater amount of
calcium is absorbed relative to the non-
nutrient alkaiine earth elements, such as strontium (Elias et
al. 1982), which resuits in a fractionation of
strontium in favor of calcium. Since calcium is one of the
principie elements used in bone and cell wall
construction, more calcium than strontium is absorbed and
deposited in bone mineral (Avioli 1988; Comar
1963; Silverberg 1990; Wasserman & Comar 1956). Although
only approximately 20% of ingested
strontium is actually absorbed into the bloodstream, 99% of the
strontium absorbed is deposited in mineral
fraction of bone (Elias et al. 1982; Hogue et al. 1961;
Kshurager et al. 1966; Lough et al. 1963; Rosenthal et
al. 1972; Samachson 1967; Schachter 1963; Spencer et al. 1973;
Schroeder et al. 1972; Taylor et al. 1962;
Wasserman 1963).
This compositional variability of bone mineral is the basic
tenet of trace element analysis, since
environmenta! exposure and dietary intake are assumed to be
responsible for such chemical alterations
(Klepinger 1984; Parker & Toots 1980). Theoretically then,
the quantity of trace elements consumed in the
diet should correlate with the quantity retained in the bone
mineral.
Since strontium intake is reflected in bone mineral, and all
mamn discriminate against strontium in favor of
calciurn during metabolisn there should be a decreased
concentration of bone strontium as one advances up
the food chain. Research indicates that strontium levels are
highest in bedrock, followed by soil, plants,
herbivores and carnivores in a given micro-habitat (Comar et al.
1957; Elias et al. 1982; Menzel & Heid
1959; Wallace & Romney 1971). Plants absorb between 50-100%
of the available strontium from the soil.
An herbivore consuming leafy matter will absorb only 20% of the
strontium it ingests relative to calcium. A
carnivore who thrives on herbivore flesh will not only ingest
less strontium than the herbivore, but will only
absorb 33% of the ingested strontium. Therefore, in examining a
particular food chain, herbivore have the
highest strontium values, carnivores the lowest, and omnivores
such as huinans, have intermediate values.
Scholars wary of this principie argue that its simplistic nature
overlooks many of the variables that make
regional, trophic-level comparisons difficult (Klepinger 1984;
Pate et al. 1991; Radosevich 1989a, 1993;
Sanford 1993; Schoeninger 1981; Silien & Kavanaugh 1982;
Sillen Smith 1984). At the base of the food
chain, soil chemistry can cause a great deal of variability in
the Sr/Ca ratios available to organisms (Pate et
al. 1991; Radosevich 1989a). For example, in many arid regions,
calcium carbonate nodules are a common
occurrence (Ezzo 1991). The calcium build-up caused by these
nodules could significantly alter soil Sr/Ca
rati and ,in turn, effect the Sr/Ca ratios throughout the
foodchain, making regional comparisons difficult.
It should not be assumed, however, that ah soil elements are
chemically mobile. Interaction among elements
depends on their availabilitY within the soil. Many factors
effect the availability of jons, including
temperature, precipitation, porosity of the soil, solubility and
weathering rates of a given minera! and soil
pH (Buol et al. 1981; Mitchehl 1957). Soil pH can have a
dramatic effect on ionic mobility such that a drop
in pH levels greathy increases the potential for ionic exchange
(Buole et al. 1981; Isermann 1981). Ionic
availability is slowly gaining in importance as a factor that
can affect both trophic level Sr/Ca ratios and the
geochemistry of archaeological bone ( Pate & Brown 1985;
Pate & Hutton 1988; Pate et al. 1989, 1991;
Radosevich 1989a,).
Plant uptake and physiology can also alter the Sr/Ca ratios of a
food chain. As Radosevich (1993) suggests,
root placement in either strontium rich or calcium-rich soil
horizons can greatly influence Sr/Ca plant ratios.
For example, in soils with a high caliche content only the roots
of trees and shrubs are massive enough to
permeate the calcareous subsurface. The Sr/Ca ratios of these
plants will be much lower than the ratios of
grassy vegetation, whose root system is not expansive enough to
penetrate the calcium-rich subsoil.
-
9
Therefore, browsing animals who rely on leaves from trees and
shrubs will have lower bone mineral Sr/Ca
ratios than grazing animals who consume various grasses.
Differences in strontium content are not solely the
result of plant species. As Isermann (1981) suggests, different
parts of the same plant (stem, leaf, frult) can
also differ in strontium content by an order of magnitude.
As discussed aboye, many variables can affect strontium
concentrations in the food chain. Variations in soil
chernistry and the chemical characterization of its parent
material, fluctuations in plant physiology and
uptake, trophic level and even animal behavior can a Sr/Ca
ratios. The role of strontium in the biosphere is a
cornplex issue and all variables must be explored before
assumptions regarding strontium distribution can
be made.
Multi-Element Research
While sorne studies focus on one or two elements, other scholars
use a multielemental approach to
reconstruct paleodiets (see reviews in Buikstra et al. 1989;
Sandford 1992; Ezzo 1992). The trace elements
are generally chosen because of their ability to distinguish
between plant a animal resources in the human
diet. For example, magnesium, manganese, strontium and vanadium
are used to infer plant consumption,
while selenium, zinc, copper and molybdenum are considered to be
indicators of animal protein (Arrhemus
1990; Beck 19 Edward et al. 1984; Edward & Benfer 1993;
Francalacci 1988, 1989; Francalacci & Tan
1988; Geidel 1981, 1982; Gilbert 1975. 1977; Hatch & Geidel
1983, 1985; Lambert et al. 1979; Liden 1990;
Morgan & Schoening 1989). Most multi-elemental studies have
focused on the shift from hunting gathering
to maize horticulturalists (Beck 1985; Buikstra et al. 1989;
Gilbert 1975; Lambert et al. 1979; Szpunar 1977;
Szpunar et al. 1978 or status based dietary differences (
Blakely & Beck 1981, Blitz 1993, 1994; Brown &
Blakely 1985; Lambert et al. 1979, 1982). The basic tenet behind
trace element research is that elemental
concentrations in bone mineral vary as a consequence of
environmental exposure and dietary intake. It is
therefore imperative to understand the association between
dietary intake and skeletal concentrations before
an element can be used to reconstruct consumption patterns.
While most scholars argue that multi-element
dietary reconstructions are more accurate than those based on a
single element, there is almost no
information concerning the distribution and function of
magnesium, manganese, zinc, seIenium copper, or
molybdenum in both pre- and post-mortum contexts (Klepinger
1984; Ezzo 1991).
Pate and associates (1991:59) include magnesium in their, list
of dietary indicators, however, swine feeding
experiments demonstrate otherwise. (Klepinger 1990). In
Klepingers research, animals fed magnesium enriched diets had bone
magnesium levels only slightly higher than those of the control
group. She argued
that if a greatly inflated magnesium intake had such a limited
affect on bone magnesium concentration, than
dietary variations found among most human groups would not be
detected by differences in the bone magnesium concentration
(1990:516).
In a thorough review of trace elements used as possible dietary
indicators, Ezzo (1991) suggests that only
barium and strontium should be used for paleodietary
reconstructions. As he rightly notes, although it is possible other
elements may be found to have potential as dietary
indicators, models based on physiological and biochemical data
have yet developed (1991:49). Considering the available data and
uncertainty surrounding many of the elements, only barium and
strontium will b used
in this research for dietary reconstruction.
While the aboye noted studies emphasize elemental distribution
in living bone mineral, an increasing
number of scholars are turning the attention towards elemental
distribution in post-mortem bone tissue
(Byrne & Paris 1987; Edward 1987; Klepinger et al. 1986;
Kyle 1986; Pate Hutton 1988; Pate et al. 1991).
Not only can diagenesis affect bone colla as discussed
previously, but it can also alter the mineral fraction
thereb obscuring the biological trace element signal.
Diagenesis in Elemental Analysis
-
10
The structural and chemical alteration of post-mortem bone
tissue is known as diagenesis. Diagenesis
primarily results m either the recrystallization of
hydroxyapatite crystals, or isoionic and heteroionic
exchange of its chemical components. The earliest
anthropological studies used strontium to reconstruct past
diet with little consideration for diagenetc processes (Brown
1973: Schoeninger 1979a, b). Over the Iast
decade, diagenesis has surged to the forefront as a process that
must be understood before elemental
analysis can be used with any degree of confidence (Price et al.
1991).
Once a bone enters a post-mortem environment, its deterioration
is mfluenced by a number of biochemical,
geochemical and organic processes. The results of these
processes are cumulative; the longer a bone remains
in a postmortem matrix, the more physical and chemical
alteration will take place (Deotare et al. 1988; Price
1989). Diagenetic changes in bone tissue generaily occur in one
of three ways (Parker & Toots 1980; Pate &
Hutton 1988; Pate et al. 1991; Radosevich 1993; Sandford 1992;
Sillen 1989): (1) Chemical components
from biological hydroxyapatite can be exchanged with soluble
jons from the surrounding soil (Ezzo 1991;
Lambert et al. 1979, 1982, 1983, 1984a, 1985b), (2) Mineral
components such as calcite (CaCO3) and barite
(BaSO4) can be precipitated to fill the tiny cracks and pores in
bone tissue (Parker & Toots 1980), and (3)
During crystallization, hydroxyapatite crystals can be altered
from biological apatite to a variety of
geochemical forms (Sillen 1989).
These processes have been grouped according to their role as
either intrinsic or extrinsic forces in bone
tissue alteration (Von Endt & Ortner 1984). Intrinsic
factors, such as porosity, density, size, and chemical
composition of bone greatly affect the degree of diagenesis. In
general, bones that are less dense, more
porous or contain a high percentage of amorphous mineral
crystais are more susceptible to diagenesis
(Grupe 1988; Newsley 1988, 1989b).
Due to their distinct degrees of calcification and density,
cortical (compact) and trabecular (cancellous) bone
should respond differently to diagenetic processes. Cortical and
trabecular bone differ markedly in their
degree of calcification; cortical bone is 80-90% (by volume)
calcified, while only 15-20% (by volume) of
trabecular bone is calcified. Consequently, the solid, cortical
bone is much denser and less susceptible to
diagenesis than trabecular bone. Trabecular bone, on the other
hand, is much more porous due to its
principie function as the mineral reservoir of the body (Baron
1990; Underwood 1977), and thus has a
greater tendency to be physically or chemically altered (Currey
1984; Price 1985; Sillen 1990)
Several studies have emphasized the differential effects of
diagenesis on bone tissue. Lambert and his
associates (Lambert et al. 1979; Szpun 1977; Szpunar et al.
1978) analyzed human remains from Woodland
sites in West Central Illinois to elucidate dietary and
biological patterning. When it became apparent that
diag was effecting elemental biological signais in the bone
tissue, their research shifted to an examination of
post-depositional processes and diagenesis (Lambert et 1982,
1983, 1984a, b, c, 1985a, b; Vlasak 1983).
In their latter research, they compared the elemental
differences between femurs and ribs from Woodland
sites in Illinois. They argued that unless diagenesis had
occurred, the dense, cortical bone of the femur
should have the same strontium levels as the trabecular rib
bone. An elemental analysis of both ribs and
femurs demonstrated that the ribs contained elevated leveis of
elements usually associated with soil
contaminants, such as potassium, iron, magnesium and aluminum.
Sodium and calcium were lower in ribs
than in femurs, while the elements assumed to be dietary
indicators (strontium, zinc and magnesium) were
found in almost equivalent levels. These results led them to
conclude that the problem with diagenesis lies
in the incorporation of extraneous elements rather than
diagenetic loss of material (Lambert et al. 1982:291).
In 1989, Buikstra and her associates re-analyzed Lamberts
research using matched rib-femur samples. While similar leveis of
zinc and lead were found in both types of bone, the ribs had a
higher concentration
of magnesium, copper, potassium, iron, manganese and aluminum
than the femora. Only calcium, strontium
and sodium were higher in the femur than in the rib. The
researchers suggest that elemental differences in
-
11
the bone pairs could be due to a greater permeability for ribs,
differential cleaning according to bone type, or
difference in bone remodeling rates.
Grupe (1988) and Brtter et al. (1977) also reported elemental
differences in cortical and trabecular
archaeological borie. Unlike Buikstra, both studies found that
trabecular bone contained a significantly
higher concentration of strontium and barium than cortical bone.
Upon examining the mass ratio of
spongiosa/compacta bone at specific locales throughout the human
body, it was determined that trabecular
bone is much more susceptible to ionic additions due to its
porous nature. In addition, trabecular bone
showed enormous variation between samples, even if they were
taken only centimeters apart.
In addition to archaeological samples, modern bone tissue has
also been analyzed for differential elemental
concentration by bone type. Tanaka et al. (1981) studied
strontium concentrations in the human skeleton of
modern Japanese populations. In support of Buikstras analysis of
archaeological remains, they found that the cancellous, rib bone
contained significantly lower levels of strontium than did the
cortical bone of the
femurs.
The available research demonstrates that differences exist in
the elemental concentrations of trabecular and
cortical bone, however, the direction of this difference is
unclear. With ah the uncertainly surrounding
elemental concentrations in clean, modern bone, archaeologists
should not use elemental differences in rib
versus femur as a method of identifying diagenesis. More
research on modern humai from the same
population consuming the same diet must be carried out before
this method can be used with any degree of
confdence. In addition as will be discussed in the following
section, there are other reliable mea for
assessing the presence of diagenesis.
While it is clear that intrrnsic factors have an important
effect on diagenesis, extrinsic factors also play a
vital role (Von Endt & Ortner 1984
Extrinsic factors such as temperature (Hare 1980;Von Endt &
Ortner 1984) groundwater (Hare 1980; White
& Hannus 1983), soil pH (Garlick 1969; Gordon & Buikstra
1981), and soil chernistry (Becker et al. 1968;
Benfer 1984; Katzenburg 1984; Keeley et al. 1977; Lambert et al.
1984a; Nelson & Sauer 1984) have all
been considered by various scholars.
A common thread running through these studies is that although
sorne elements, such as iron, manganese,
aluminum, and potassium are more susceptible to postmortem
alterations than others (strontium, lead and
zinc), no one element is immune from diagenetic processes. For
example, strontium, once thought to be
generaily stable in post-depositional environments (Parker &
Toots 1970, 1980; Pate & Brown 1985), can
be highly mobile under certain geologic, geographic and temporal
conditions (Nelson et al. 1986; Pate et al.
1991; Radosevich 1989a, b, 1993).
Pate and his associates (Pate & Hutton 1988; Pate et al.
1989) attempted to develop a model of ionic
exchange between bone and the surrounding soil matrix. They
focused on the solubility of ions m arid soils
of South Australia and their presence in archaeological bone.
Analyzing for a suite of elements, they found
that archaeological bone was depleted in magnesium, yet enriched
in aluminum, iron, barium, strontium,
potassium and manganese. Only zinc and sodium were found m near
equal quantities in both bone and soil.
Their results suggest that as bone deteriorates, the dissolution
of apatite crystals permits the incorporation of
carbonates and other ions into the crystailine structure. It was
also demonstrated that there is an association
between collagen degradation and degree of diagenesis; as
collagen decomposes the physical and chemical
alteration of the mineral fraction will also increase. When
microbes decompose collagen, they form acidic
by-products that accelerate the rate of ionic exchange in the
hydroxyapatite crystal (White & Hannus 1983).
Pate and his colleagues found that the enrichment of aluminum,
iron and potassium was probably due in part
to soil contaminants that were incorporated into the mineral
fraction, but not completely removed before
analysis.
-
12
Price (1989) studied the effects of diagenesis on bone samples
from Skateholm, Sweden, Nelson Bay Cave
in South Africa, and the Price] site in Wisconsin. Utilizing
multi-elemental analysis, he found an association
between degree of diagenesis and age of sites in the Skatel
sample. The older samples from the Skateholm II
site, had lower bon calcium and collagen levels than did the
those from the younger Skateholm 1 site. In
addition, sodium, strontium, copper, magnesiuni manganese and
aluminum were much more abundant in the
older samples than the younger. The lower calcium and collagen
leveis at t older site indicate a greater
degree of bone degradation and diagenesis. The reduced calcium
levels also appear to make the other
elements loi more abundant in proportion, which account for the
high leveis of the elements in the older
samples. The strong positive correlation betwee strontium and
iron, and between magnesium, aluminum and
iron suggest that many of the samples contained soil components
that were incorporated into the bone
mineral. During diagenesis, bone calcium was replaced by soil
ions, thereby reducing calcium levels and
increasing th quantity of magnesium and alummum.
At Nelson Bay Cave, Price found calcium values fairly stable
over ti with an average of 35.7% (suggesting
good preservation). There was a very low correlation between
calcium and the other elements analyzed
(aluminum, copper, iron, magnesium, manganese, sodium,
strontium, and zinc). The strongest positive
correlations occurred between iron an aluminum, and sodium and
magnesium. The similar behavioral
pattern of rnagnesiurn, iron and aluminum probably suggest the
presence of soil contaminants. The elevated
sodium levels in the older samples suggest incorporation of
these ions during apatite recrystallization, which
increases with length of deposition.
The data from the Price III site show that calcium values
decreased with the sample age. The only
significant positive correlation was between iron and aluminum,
while strontium was not strongly correlated
with any other elernent. A strong negative correlation was
fornid between calcium and manganese. The
decreasing calcium values with sample age suggests a diagenetic
loss of calciurn over time. Since
manganese is h mobile in post-depositional environments (Lambert
et al. 1984a), its negative correlation
with calcium supports the idea of diagenesis through ionic
exchange. The positive correlation between iron
and aluminum suggests bone mineral enrichment by soil
contaminants.
Prices research dernonstrates that while diagenesis is prevalent
in archaeological bone, the forrn and degree of diagenetic
alterations are dependent upon geologic, geographic and temporal
conditions. Before dietary
patternmg can be inferred, multi correlations should be
conducted to assess the degree of diagenesis.
Elemental correlations can indicate bone mineral enrichment by
soil contammants, apatite recrystallization,
ionic exchange, or bone deterioration.
While the problem of diagenesis has been considered by many
scholars in their research, only now are sorne
devising methodologies to control for its affects. Lambert and
co-workers (1989, 1990, 1991), have experimented with sample
cleaning procedures. Their methods for obtaining elemental
biogenetic
concentrations include: (1) abrading and washing the sample in
distilled water, (2) physically removing 1-3
mm the outer bone cortex, (3) treating the powdered bone with 1N
acetic acid, and (4) repetitively washing
the powdered bone in an acetate buffer.
Since each method appears to be beneficial in obtaming the
biogenetic signal of specific elements, the
authors endorse a combined approach in processing bone samples
for analysis. They also contend that
element such as magnesium, potassium, sodium, and zinc can be
used as dietaz indicators when bone is
processed correctly (Lambert et al. 1991). Washing powdered bone
in an acetate buffer restores original dietary zinc levels, while
the removal of the outer bone cortex removes most diagenetic
effects of
potassium, magnesium, and sodium.
A number of other washing procedures have been developed to
control for diagenesis (Burton & Price
1990a, 1990b; Ezzo 1991, 1992; Fn et al. 1991; Sillen 1986,
1989; Sillen and LeGeros 1991). Sillens procedure calls for washing
powdered bone in an acetate buffer with a pH of 4.5. The solution
is
-
13
centrifuged, the supernatent is decanted and saved for elemental
analysis, and more buffered acid is added to
the residue. This sequence is then repeated another 20-25 times,
so as to create a solubilil profile (Sillen 1989). Siliens
methodology was ground-breaking in that it demonstrated that Ca/P
ratios of archaeological bone could resemble those of modern bone
by means of a sequential buffered acid wash. Although his
methodology holds great promise for diagenetic research, his
procedure (which calis for 20-25 supernatants
per sample), is both labor intensive and time-consuming. In
addition, Siliens methodology analyzes the liquid decanted from the
sample rather than the digested bone residue.
Price and associates (Burton & Price 1990a,1990b: Price et
al. 1991) improved upon Siliens methodology in several ways. They
wash whole bone in 1N unbuffered acetic acid, ash (to remove all
organics), grind to
a powder, and digest in nitric acid. Unlike Siliens method, in
which the bone powder was centrifuged and removed before analysis,
Price and Burtons method analyzes the bone in solution Using this
procedure, they demonstrated that archaeological bone Ca/P ratios
could be reduced to modern levels (approx. 2.1),
thereby eliminating most diagenetic affects and recovering
biogenetic signals. Perhaps the most important
improvement is the speed with which the samples can be processed
and analyzed. Since the powdered bone
ash is digested only once, as opposed to the 20 times as in the
procedures of Sillen, elemental analysis can
be done with much less investment of time and labor.
In addition to washing methods, a number of other methods can be
used to assess the nature and degree of
diagenetic change m bone tissue. As mentioned aboye, the ratio
of calcium to phosphorus m archaeological
bone is a good indicator of post-mortem contamination. Since
both calcium and phosphorus function
primarily as the crucial components for bone mjneralizafion,
their quantities are constant in bone tissue
(Underwood 1977).
Phosphorus accounts for approximately 17% of the mass of modern
bone mineral, while calcium levels
approach 37% (Avioli 1990; Broac 1990; Woodard 1962). Therefore,
as Katzenberg (1984) and
Sillen(1981b) suggest, archaeological bone, free of diagenesis,
should have the same Ca/P ratio as that
found in modern bone (approximately 2.15). Ratios deviating from
this value suggest either enrichment or
depletion of calcium and/or phosphorus. For example, when
calcium is depleted from bone mineral, the
Ca/P ratio is lower than 2.15, and the other element may appear
to be present in higher quantities. Because
spectrographic analysis measures the amount of one element in
relation to all other elements in the sample,
if one element, such as calcium, is low, the other elements may
appear high in proportion. A Ca/P ratio
higher than 2.15 can occur when bone mineral is enriched in Ca
during diagenesis.
The percentage of ash content of archaeological bone also
provides a measure of diagenetic change. Living
bone consists of approximately 70% inorganic mineral distributed
throughout an organic matrx (Underwo
1977). When ashed, clean, modern bone will have about 70% (by
weight) remaining as a mineral fraction.
To assess the integrity of archaeological bone, then, the bone
ash % should be very near to 70%. If the bone
ash % is lower, it may suggest that diagenesis is removing
mineral components from the bone. If the bone
ash % is higher than 70%, it can indicate one of two processes:
(1) an enrichment of bone mineral by
anionic or cationic substitution from the surrounding matrix, or
(2) the deterioration of the organic portion,
which would cause the minera! portion to appear greater in
quantity.
Another method that has been used successfully to identify
diagenesis is x-ray diffraction (Kyle 1986;
Piepenbrink 1986; Schoeninger 1981, 1982; Schoeninger et al.
1989; Sillen 1981a, b, 1989). This
methodology is based on the principie that normal, unaltered
bone mineral has a diffraction pattern similar
to apatite. Thus, geological hydroxyapatite (with a different
chemical composition) can be distinguished
from biological hydroxyapatite crystais on the basis of
differing diffraction patterns. When archaeological
bone displays patterns that vary from expected biological
apatite, such as the presence of large amounts of
calcium carbonate, diagenesis or recrystallization is assumed to
have taken place.
-
14
It has also been suggested that archaeological fauna! remains be
used as a guide or baseline for identifying
human elemental values that extend far beyond the expected range
(Brown 1973; Katzenberg 1984; Price
1985; Price et al. 1985a; Schoeninger 1979a, b, 1981, 1982;
Sillen 1981). While both carnivores and
herbivores are equally important in defining a range for human
values, carnivores have been used much less
frequently (Radosevich 1993). The limitation of using carnivores
stems from their variable diets, which can
be supplemented with bone and plant materials (Katzenberg 1984;
Schoeninger 1981, 1982), and their
scarcity from archaeological sites. Sorne scholars are instead,
advocating the use of herbivores since their
diets are assumed to be homogeneous and relatively stable over
time, even in the face of human interaction
(Elias et al. 1982).
A number of other techniques have been used to identify and
assess the effects of diagenesis. These include
the analysis of archaeological and their surrounding soil
(Becker et al. 1968; Benfer 1984; Katzenber 1984;
Keeley et al. 1977; Lambert et al. 1984a; Nelson & Sauer
1984), the comparisons of trabecular to cortical
bone (Lambert et al. 1982), the comparison of bone tissue to
dental tissue (Parker & Toots 1970, 1974)
comparison between prehistoric populations and modern equivalent
(Edward et al. 1984; Hancock et al.
1987, 1989; Gilbert 1975; Lambert et 1985a; Parker & Toots
1970, 1974, 1980; Szpunar 1978), and
microprobe analysis (Gilbert 1975;Hassan & Or 1977; Hende et
al. 1983; Lambert et al. 1983, 1984a,b,c,;
Schoeninger 1979a, b,; Waldron 1981; Vlasak 1983).
Diagenesis versus Diet at Monte Albn
In an attempt to elucidate inter-elemental relationships, a
correlation matrix was constructed for the log-
normal values (Table 3.2). The values prmted in boid represent
correlation coefficient that are significant at
p less than or equal to 0.05. These statistically significant
correlations raise interesting questions about the
nature of bone and human diet, and will be addressed below.
TABLA QUE BORR 3.2 Table 3.2. Correlation Matrix for
Multi-Elemental Values from Monte Albn
One of the most numerous inter-elemental correlations involves
aluminum (Al). As one on the most
common metais in the earths crust (Hammond 1975), aluminum is
often an indicator of a soil contammant when correlated with
potassiuni (K), iron (Fe) and manganese (Mn) (Ezzo 1992; Lambert et
al. 1979,
1985a). As Table 3.2 illustrates, the strongest correlatjons are
between Al, Fe, K, Mn and Zn. The
correlation between Al and Ba is perhaps evidence of oxide
contaxnination, which often results in a
correlation between these 2 elements (Parker and Toots 1972). I
would also like to point out that a
correlation between Al and K does not necessarily signal
contamination or diagenesis. Table 3.3 shows the
results of a correlation matrix of 20 samples from modern cow
bone (elemental values usted in Appendix C)
Even in a clean, modern bone, free of contaminants, there is
still a atatiscally signicant correlation between
Al and K
Table 3.3. Correlation Matrix for Modern Cow Samples
The strongest correlation in the Monte Albn matrix is between Ca
and P. This is expected since these
elements are the major components of bone, constituting 37% and
18% of bone respectively. The linear
relationship between Ba and Sr is reflected in Table 3.2 as the
strongest correlation that either Ba or Sr have
with any other element. The positive correlation suggests that
because the 2 alkaline earth elements mirror
on another in behavior, the data set have not been substantially
altered by diagenesis.
Diagenesis is a problem that cannot be avoided. It affects all
buried bone to sorne degree, and in certain
circumstances, can completely alter the dietary signal of the
bone. As discussed in Chapter 4, several
-
15
methods have been used in this research to test for and reduce
the impact of diagenetic alteration of the
mineral fraction. Since bone ash (which is largely mineral in
composition) is approximately 67% of whole
dry bone, the elevated ash% values usted for the Monte Albn
samples (Appendix E) seem to indicate an
addition to the mineral fraction via diagenesis. As Table 3.2
illustrates, however, although there is a
correlation between several soil elements, the presence of a
statisti significant correlation between Sr and Ba
demonstrates that the dietary signal is still retained. Perhaps
the high ash% values do not indicate an
addition to the mineral fraction, but rather a depletion of the
organic fraction over time. Adding strength to
this hypothesis is that during preparation for isotopic
analysis, only 40 of the original 50 samples had
enough collagen to analyze. If 20% of the samp!e lost almost all
their organic fraction, than it can be
assumed that the other 80% lost sorne of their organic fraction
as well. When ah unes of evidence are
examined, although the Monte Albn samples may have been affected
by diagenetic processes, the mineral
portion is reflectjve of diet and can be used in reconstructing
past behavior.
While alrnost all of the eleven trace elements usted have been
used in the reconstruction of ancient diet
(Buikstra et al. 1989; Klepinger 1990), only Sr, and to a lesser
extent Ba, have been shown to be reliable
indicators of paleodiet (Burton and Price 1990a, b; Elias et al.
1982; Ezzo 1991; Weydert 1990). Therefore,
due to the uncertainty surrounding the utility of the trace
elements, 1 will emphasize only Ba and Sr in this
research purposes of dietary reconstruction.
Summary
Reconstructing diet on the basis of isotopic ratios and
quantities of trace elements began almost 20 years
ago. During that time, carbon isotopes strontium were used
alrnost exclusively to identify dietary
components in human diet. Advances in technology and an increase
in laboratory research suggested that
nitrogen isotopes and a number of other trace elements could be
used as dietary indicators. While mtrogen
can be used to determine a marine component in the diet, its
ability to detect meat and legume consumption
is still unclear. Another innovation in isotopic analysis
involves the comparison of bone collagen and bone
apatite ratios to determine specific dietary components. While
the role of lipids, carbohydrates and protein
in the synthesis of collagen and apatite carbon is still
unclear, new research by Ambrose and Norr (in
preparation, as cited in Ambrose 1993) suggest that the 13C
value of apatite reflects the diet as a whole,
while the 13C collagen reflects only the protein cornponent.
Consequently, a comparison of collagen and
apatite 13C values may provide some of the data needed to
distinguish different dietary fractions among
ancient populations. Innovations in trace element analysis
stressed the utility of nurnerous elements in
reconstructing the importance of specific food items to the diet
(i.e. magnesium for nut consumption).
Unfortunately, of all the assumptions surrounding various trace
elements, only barium and strontium have
been found useful as indicators of past diet.
Along with research on new dietary indicators carne an interest
in the post-mortum contamination of
archaeological bone (diagenesis). An increased awareness of
diagenesis spurred numerous advances in the
analytical techniques used for its detection, and the sample
preparation methods used to lessen its
destructive forces. Although all archaeological bone is
chemically or physically altered over time (Badone
& Farquhar 1980; Sandford 1993), the presence of diagenesis
does not impede the ability to obtain
paleodietary information from the sample. The use of multiple
methodological and analytical strateg such as
acid washing faunal and soil analysis, Ca/P ratios, and
multi-elemental analysis, have been used to
determine diagenetic processes and recover biogenetic signals.
In order to successfully test for diagenesis
and conduct meaningful isotopic and elemental analyses of the
Monte Albn skeletal population, a set of
procedures for the selection and preparation of the samples was
devised. The methodology and reasoning
behind these procedures are presented in the foliowing
chapter.
Notes:
-
16
1. Prior to taking the naperian logarithm of the entire data
set, all negative Fe and Mg values had to be
eliminated (see Appendix D). Using Buikstra et al.s (1989)
method, all elemental values were increased by 100 and then
multiplied by 10. Once these mathematical procedures were
completed, the values were
logarithmically transformed and a correlation matrix
constructed
Sample Selection and Preparation
Introduction
For isotopic and elemental analysis to yield meaningful results,
the role of diagenesis and the original
research question must be considered when selecting and
procuring samples for analysis. To this end, 1
applied four separate criteria when selecting the Monte Albn
sample. To limit the affects of diagenesis,
samples were selected so that variability in bone preservation
and skeletal part were kept to a minimum.
Moreover, sinsce the primary focus of th.is research rests on
the question of social inequality and dietary
patterning, samples were selected on the basis of interment type
and provenience. By selecting samples
from a continuum of burial contexts (tomb to shallow grave), and
locations (site core to periphery), it was
assumed that a more representative sample of social status
levels would be collected. Once the samples were
chosen, they were prepared the laboratory for both isotopic and
elemental analysis. The following chapter
describes the criteria used for sample selection and the
procedure followed for sample preparation.
The Monte Albn Sample
Background
The Monte Albn burial population was sampled during June and
July 1990, and June 1991, at its two
curated locations; the Laboratory of Physical Anthropology in
the National Museum of Anthropology and
History, Mexico City, and the store rooms of the historic
ex-monastery in Cuilapan, Oaxaca. The samples
from Mexico City were collected under the auspices of Sergio
Alonso Lopez, then Director of the
Department of Fhysical Anthropology. The samples from Cuilapan
were collected under the direction of
Ernesto Gonzalez Licn, then Director of Archaeology at Instituto
Nacional de Arqueologia e Historia
(INAH), Oaxaca City, and Lourdes Marquz Mrfin, then director and
curator of Physical Anthropology
also at INAH, Oaxaca City. Before sampling was conducted, a list
of many of the burials at Monte Albn by
age and sex was provided to me by Dr. Richard Wilkinson of the
State University of New York at Buffalo.
A complete list of the burials recovered from Winters (1972)
excavation of Monte Albn also was provided by Dr. Marc Winter of
INAH, Oaxaca City, the principal investigator, and Dr. Richard
Wilkinson, the
physical anthropologist associated with the project.
Skeletal Part and Preservation
Bone samples were selected for analysis on the basis of body
part and age of individual. As discussed m
Chapter 3, elemental concentrations vary widely throughout the
human skeleton. Research demonstrates that
intra-skeletal concenfratio have a greater amount of variability
than do same bone comparisons between
individuals with similar diets (Brtter et al. 1977). The
tremendous amount of variability within one
individual is due to the different physical characteristics of
trabecular and cortical bone.
Cortical bone, which is dense and compact, interacts much
differently with trace elements than does porous,
trabecular bone. One means to use. to limit the natural
variability found throughout the skeleton is to select
the same skeletal part throughout the analysis. Research
conducted by Grupe (1988) suggests that elemental
concentrations from the shafts of long bones represent the
average elemental concentration of the entire
skeleton. In this research, therefore, I collected only long
bone shafts for analysis. Initially only mid-shaft
sections of the femora were collected. However, in some
instances the femur was either in a complete state
and could not be sampled, or was missing, and so the tibia was
instead sampled. If both the femur and tibia
were unavailable, the humerus midshaft was collected. In several
instances, the skeleton was so incomplete
-
17
that other bones were sampled. Table 4.1 provides a summary of
the skeletal elements sampled from the
Monte Albn burial population. Por a more complete listing, see
Appendix A.
Approximately one gram of bone was removed from each long bone
shaft by means of a small, notched-
tooth saw. Each bone was placed in plastic bag and labeled by
burial number. A paper tag was placed
around the remaining cut bone shaft, indicating that it had been
artificially altered for the purposes of
chemical analysis. A list was made of all samples collected and
placed on file with the Departments of
Physical Anthropology at the National Museum of Anthropology and
History, Mexico City, and at INAH,
Oaxaca City.
Table 4.1. Skeletal Elements Sampled
Element n
Femur 152
Tibia 39
Humerus 24
Fibula 10
Ulna (cubito) 1
Radius 1
TOTAL 227
Ah sample collection was conducted under the supervision of Jos
Jimenz Lpez in Mexico City, and
Lourdes Marquz Mrfin in Oaxaca City. These physical
anthropologists also supplied information
regarding the sex and age of specific individuals when these
data were missing from the master lists. For the
purpose of this research, only adults (over 15 years of age)
were selected for analysis, since elemental
values taken from children fluctuate widely and inconsistently
(Price 1989). Table 4.2 ihlustrates the
breakdown of the Monte Albn sample by sex. For a more detailed
account, see Appendices A and B.
Table 4.2. Monte Albn Sample by Sex TOTAL 227
Interment Type
The individuals collected for a were recovered from several
different burial contexts. Those sampled from
Caso and Acostas excavations at Monte Albn were labeled with a
specific nomenclature according to interment type (Romero 1942).
The term tomb referr an interment consistmg of one or more
underground chambers.
According to Romero, the individual occupying the primary chamb
tomb was considered the principal
occupant, and labeled T.l through T.172 according to tomb
number.
The term entierro (interment) had a very generalized usage, and
referred to any individual who was not
buried as the principal occupant a tomb. Thus, entierros were
found in simple, shallow graves on the slopes
of the site, in stone-lined graves on the Main Plaza, or as
secodary burials of a tomb. Usually, the labels for
entierros were more specific than tombs, indicating both
entierro number and fleid season. For example in
the label Ent. IX-3, the roman numeral stands for the ninth
field season, and the 3 represents the third entierro found during
that field season.
Sorne confusion arises, however, when several individuals were
recovered from a single tornb. The tombs principal occupant is
given the T.153 !abel, while the other occupants (found in minor
rooms, niches, or dissarticulated in the main chamber) were labeled
as entierros within a tornb, such as T.153, ent. 3.
-
18
Several tombs contained a large mass of disarticulated bones in
which single individuals could not be
identified, and so the area was divided and exacavated in
quadrants (1-1V). ui these instances, the bones
were labeled according to their quadrant (e.g. T.128, ent. sec.
II) Since very little information is available regarding the
entierros, it is not known if the entierros in tombs are temporally
related to the principal
occupant, or if they were once principal occupants themselves,
moved as a consequence of tomb re-use.
When Winter conducted his excavation of Monte Albn, he chose to
use a new method of labeling
interments (Winter 1974; Winter and Payne 1976; Winter and
Wilkenson n.d.). Whereas the tomb
designation was still used, Winter chose to use burial rather
than entierro. In addition to the burial designation, Winter also
assigned either house number or individual number, and the year of
excavation to
each interment (e.g. Burial 25, M.A.1973B, House 3 or Burial 14,
ind.1, M.A. 1972).
Many hours were spent deciphering the association between
interment classjficatjon and actual burial
context. Since Wilkinson and Norelli (1981) used only a tomb/non
tomb division to assign social status to the skeletal population at
Monte Albn, I thought it necessary that the bur context of all
interments
(entierros and tombs) be clearly defined befon status
assumptions were posited. As suggested aboye, the
term entierr should not be used to identify social status (as
Wilkinson and Norelli c Entierros can be
inhabitants of shallow graves, but they can also be fou in
elaborate tombspossibly representing past principal occupants.
Consequently, in this research, the actual burial context,
rather than the interment classification became the
important variable in selecting bone samples for analysis.
Interment Provenience
Burial location was also selected as a variable in determining
social differentiation at Monte Albn
(Appendix 1). During Blantons survey of the area, he discovered
15 subdivisions of the site, each with its own cluster of elite
residences and civic-ceremonial structures. He refered to these
areas as barrios and
numbered them from 1 to 15 (Blanton 1978) (Figure 4.1). Due to
the differential distribution of status-
related artifacts such as exotic rocks and minerals, marine
shell, and exotic pottery type Blanton suggested
that several of the barrios closest to the Main Plaza had a
stronger elite presence than barrios farther
downslope. Assuining a possible status differentiation by
barrio, I attempted to assign each skeletat sample
to its appropriate barrio for analysis.
A thorough review of ah the published and unpublished data
available was necessary to determine the
spatial distribution of the Monte Albn sample. Alrnost all of
the tombs were located in the Main Plaza or
just north in an area known as the North Cemetery (see Figure
4.1). These spatial data were collected from a
variety of sources, including the published reports, letters and
manuscripts of Caso and his associates (Caso
1934, 1935, 1937a, b), and the field notes and maps from
Blantons survey of Monte Albn (1978). As illustrated in Appendices
A and B, almost ah tornbs have been located according to the
numbered terrace it
rests on, and the barrio in which it is found.
Caso and his fellow researchers placed a heavy emphasis on the
elaborate tornbs of Monte Albn, while
making only cursory references to the entierros in rnost
published articles and unpublished manuscripts.
Sorne of the entierros could be assigned to a specific terrace
due in large part to the careful and detailed
field notes and maps of Blantons survey of the Monte Albn area
(1978). When the specific locale of an entierro could not be
obtained from the available data, a general barrio location was
assigned (Appendices A
and B). Barrios were assigned on the basis of where excavations
were conducted during a particular field
season. For example, during the second season of field work,
Caso and associates excavated in and around
the Mam Plaza (Caso 1932). Although this area contams several
different terraces, they are all located in
barrio 2. Thus, aH entierros and tombs recovered during
Temporada II were placed in barrio 2. In this way, a majority of
entierros whose provenience was previously unknown could be given a
general barrio
affiliation, thereby expanding the scope of the skeletal
population across space. Of the 83 entierros available
-
19
for analysis, 5 were securely located to barrio, 23 had barrio
Iocations that could be inferred, and 55 samples
had no provenience data present.
.
A majority of the samples come from barrio 2 (Table 4.3). The
apparent emphasis on this barrio stems from
the numerous excavation conducted by Caso and his associates in
and around the structures of the Main
Plaza. During their 18 field seasons at the site, most of their
energy was focused on the recovery and
reconstruction of the buildings encircling the Main Plaza, and
the associated burial tombs and chambers.
The only excavations outside of barrio 2 were conducted during
Temporadas 1V, VIII and XII, when
interments were recovered from barrios 1, 4, 6, and 8 (Acosta n.
d.; Caso 1935, 1942).
Following the last excavation in 1958 the site stood relatively
untouched until 1971, when Richard Blanton
began his settlement pattern study of Monte Albn (Blanton 1976a,
1976b, 1978; Neeley 1972). In 1972 and
1973, Marc Winter carried out an excavation downslope from the
Main Plaza, in what is considered to be
the residential zone of the city (Winter 1974; Winter and Payne
1976). He uncovered several small
residential units in barrio 7, and a number of human burials.
With the help of Richard Wilkinson, Winter
recorded the sex, age, burial type and associated grave
furniture for each individual (Winter & Wilkenson
n.d.) On the basis of burial information and residential type,
size and complexity, Winter suggested that
these individuals represented the middle or low social segment
of society at Monte Albn (Winter 1974;
Winter and Payne 1976).
Sample Preparation and Analytical Procedure
Trace Element Analysis
The preparation and analysis of all samples were done in the
Laboratory for Archaeological Chemistry,
Department of Anthropology, at the University of
Wisconsin-Madison. Prior to analysis, the samples were
cleaned usmg a variety of methods adapted from Siliens research
(1986, 1989). Approximately 1-3 mm of outer bone cortex were
mechanically removed from each sample by means of a hand-held
sander with an
abrasive bit. Each sample was placed in a polyethylene vial
containing 1N acetic acid and sonicated
(cleaned via somc waves) for 30 minutes.
According to SilIen (1986, 1989), the acid wash is necessary to
remove diagenetic carbonates from the
mineral fraction, which can cause an artificial elevation of the
calcium and strontium values. After
sonication, the samples were rinsed repeatedly in deionized
water and transferred to a vacuum chamber,
where more 1N acetic acid was added. By using a vacuum chamber,
acetic acid comes in contact with the
entire surface area of the bone, since air is drawn out of the
bones microscopic pore space in exchange for acetic acid. All
samples were left in vacuum for 2 hours, after which time they were
left to soak in solution
for approximately 12 hours. Next, the samples were thoroughly
rinsed in deionized water and placed in a
drying oven at 80C for 24 hours.
Once the samples had been cleaned and dried, the dry weight of
each sample was recorded. They were then
placed m a porcelam crucible and reduced to ash in a muffle
furnace for 8 hours at a temperature of 725C.
Once cooled, an ash weight was recorded, and the remaining ash
percentage was calculated (Appendix E).
Next, each sample was ground to a fine powder and approximately
50 mg of this powdered bone ash was
placed in a test tube.
Digestion proceeded when 1 cc of concentrated nitric acid (HNO3)
was added to each test tube and heated to
110C for 1 hour. Once cooled, 9.5 cc of 5% nitric acid was added
to all samples. Following digestion, the
samples were analyzed for eleven elements (alummum, barium,
calcium, iron, magnesium, manganese,
sodium, phosphorus, strontium, and zinc) by means of inductively
coupled plasma emisson spectroscopy.
In addition, to the 227 Monte Albn samples analyzed, 4 bone
standards were periodically analyzed to
-
20
assess the machines precision. One standard used was the
Internal Atomic Energy Comrnission bone standard, H5. However,
since the quantity of H5 was limited, three other internal
laboroatory standards
(B0126, B5407 and CowRef) were used in conjunction with H5. The
results of the elemental analysis of the
Monte Albn samples are provided in Appendix A, while those of
the bone reference, H5, are provided in
Appendix F.
Stable Isotope Analysis
A subset of 40 individuals was chosen for stable carbon and
nitrogen isotope analysis. These individuals
were selected on the basis of thei, associated time period,
since my research focuses on temporal shifts in
dietary behavior. Stratified random sampling was used to select
the individuals for analysis. Since this
research aims to explore the diachronic nature of social
inequality via dietary reconstruction, the samples
were divided by time period (MA II, IIIa, IIIb, IV, and V) and
samples were then randomly chosen from
each time period. Equal numbers of individuals from time period
was the ideal, however, due to the limited
availability of samples from sorne time periods, and badly
preserved collagen from others, not ah time
periods are respresented in the isotope subset. Additionally,
while it was hoped that ah barrios would be
represented, only samples from barrios 2, 6, 7, and 8 were
available for analysis. Table 4.4 illustrates the
number of samples analyzed for each time period and barrio.
Sample preparation and analysis were done
under the direction of Margaret Schoeninger, in the Paleodiet
Laboratory, Department of Anthropology at
the University of Wisconsin-Madison. A variety of collagen
contaminants can affect isotope ratios. Lipids,
post-depositional carbonates, biological carbonates in bone
apatite, and even soil particles and other organic
matter adhering to the bone cam alter isotopic composition
(Hassan & Ortner 1977; Hassan et al. 1977;
Hanson & Buikstra 1987; Kyle 1986; Piepenbrink 1986). In an
attempt to control for these contaminants, all
samples were mechanically and chemically cleaned prior to
analysis.
Approximately 1 g of bone was selected from which the first 1-3
mm outer cortex was removed to rid the
sample of any adhering contaminants. This was followed by an
ultra-somc cleaning in distilled water for 30
minutes, after which time they were placed in a lyophilizer for
16 hours at a temperature of 80C.
Ah samples were placed in beakers containing .1M hydrochloric
acid (HCI) for a period of approximately 2
weeks. The samples were checked daily and the acid was changed
on a weekly basis. The use of a weak (1-
3%) hydrochloric acid not only removes the mineral bone fraction
and any post-depositional carbonates, but
is better than strong acids (10%) at recovering collagen from
poorly preserved samples (Schoerunger et al.
1989). Once the samples were demineralized, the collagen was
rinsed to neutrality in deionized water and
left to soak in .125M sodium hydroxid (NaOH) for approximately
12 hours.
Treatment with sodium hydroxide removes any humic acid and lipid
contamjna that can affect collagen
carbon isotope ratios. Whereas the isotopic composition of humic
acid reflects the local plants rather than
diet, bone lipid isotope ratios can be 6-12% more negative than
bone Collagen ratios (DeNiro and Epstein
1977; Ambrose 1990).
Upon rinsing in deionized water, the samples were dried in a
lyophilizer for approximately 16 hours and a
dry weight was recorded. A percent weight of collagen was
calculated for each sample; those with values
aboye 5% were considered acceptable for analysis, while those
wit a collagen percentage below 5% were of
questionable composition and ha to be analyzed further using
atomic C:N ratios (Schoeninger et al. 1989).
As research by DeNiro (1985) illustrates, reliable collagen
samples use for isotopic analysis typically have
atomic ratios ranging from 2.9 to 3.6. Ratios higher than 3.6
can indicate collagen contamination by humic
acids, lipids, or carbonates, and should be rejected for
analysis. As shown in Appendix G, of the forty-two
samples analyzed for carbon and nitroger isotopes, three had a
C:N ratio higher than 3.6 and were
subsequently exciuded from this research.
Following collagen purification, the samples were completely
converted to C02 and N2 gases via static
combustion (Sofer 1980). Between 3 and 5 mg of collagen were
weighed into silver boats and loaded into a
-
21
heated quartz tube with copper oxide and diatomaceous earth. The
quartz tubes were then placed under high
vacuum for several hours to remove all moisture, sealed with a
torch and then placed m a furnace and heated