Martu ethnoarchaeology: Foraging ecology and the marginal value of site structure

Abstract

Archaeological investigations of hunter-gatherer site structure have remained largely descriptive, despite significant explanatory advancements by evolutionary approaches to foraging behavior and ecology. To date, calls to incorporate site structure studies within this behavioral ecological framework have largely been ignored. We suggest there is a clear explanation for this. At large spatial extents, human behavior is constrained by patterned ecological variability, as such, a general theory of behavior is likely to characterize key aspects of human decisions. At small spatial extents, human behavior is not constrained by patterned ecological variability, therefore, the human decisions that produce site structure should be driven by mechanical constraints or random variation. However, variation in site structure may be ecologically relevant inasmuch as it informs on landscape level variation in human-environment interactions. Drawing on ethnoarchaeological data collected in collaboration with Martu, Aboriginal foragers in Western Australia, here we test empirically-derived, mechanistic predictions on site size and material size sorting to show how these can inform theoretically-derived, adaptive predictions from the Marginal Value Theorem. Results show that site size increases with the number of occupants and hence, the amount of in-patch foraging competition, while size sorting increases with the duration of occupation and hence, in-patch residence time. Combined, these attributes of site structure can be used as proxies of foraging behavior to explain variability in overall foraging yields. With this approach, site structure can provide insights into foraging decisions that can be examined through a general theory of behavior.

Full text

UNCORRECTED PROOF
ARTICLE INFO
Article history:
Received 12 January 2016
Received in revised form 20 June 2016
Available online xxx
Keywords:
Hunter-gatherers
Site structure
Marginal value theorem
Behavioral ecology
ABSTRACT
Archaeological investigations of hunter-gatherer site structure have remained largely descriptive, despite significant ex-
planatory advancements by evolutionary approaches to foraging behavior and ecology. To date, calls to incorporate site
structure studies within this behavioral ecological framework have largely been ignored. We suggest there is aclear expla-
nation for this. At large spatial extents, human behavior is constrained by patterned ecological variability, as such, a gen-
eral theory of behavior is likely to characterize key aspects of human decisions. At small spatial extents, human behavior
is not constrained by patterned ecological variability, therefore, the human decisions that produce site structure should be
driven by mechanical constraints or random variation. However, variation in site structure may be ecologically relevant
inasmuch as it informs on landscape level variation in human-environment interactions. Drawing on ethnoarchaeological
data collected in collaboration with Martu, Aboriginal foragers in Western Australia, here we test empirically-derived,
mechanistic predictions on site size and material size sorting to show how these can inform theoretically-derived, adap-
tive predictions from the Marginal Value Theorem. Results show that site size increases with the number of occupants
and hence, the amount of in-patch foraging competition, while size sorting increases with the duration of occupation and
hence, in-patch residence time. Combined, these attributes of site structure can be used as proxies of foraging behavior
to explain variability in overall foraging yields. With this approach, site structure can provide insights into foraging deci-
sions that can be examined through a general theory of behavior.
© 2016 Published by Elsevier Ltd.
Journal of Anthropological Archaeology xxx (2016) xxx-xxx
Contents lists available at ScienceDirect
Journal of Anthropological Archaeology
journal homepage: www.elsevier.com
Martu ethnoarchaeology: Foraging ecology and the marginal value of site structure
Brian F. Codding a, ⁎, David W. Zeanah b, Rebecca Bliege Bird c, Christopher H. Parker a, Douglas W. Bird c
aDepartment of Anthropology and Archaeological Center, University of Utah, United States
bDepartment of Anthropology, Sacramento State University, United States
cDepartment of Anthropology, Penn State University, United States
1. Introduction
Ecological and evolutionary approaches in hunter-gatherer archae-
ology continue to make tremendous strides toward explaining vari-
ation in subsistence (e.g., Broughton et al., 2011; Speth, 2010), set-
tlement (e.g., Byrd et al., 2015; Winterhalder et al., 2010), technol-
ogy (e.g., Surovell, 2012; Stevens and McElreath, 2015), storage (e.g.,
Morgan, 2012; Whelan et al., 2013), and demography (e.g., Kelly et
al., 2013; Williams et al., 2015). But despite this progress, studies of
site structure remain largely descriptive (e.g., Hill et al., 2011; Speth
et al., 2012).
O’Connell (1995) made this same observation twenty years ago,
noting the disparity between ethnoarchaeological studies of faunal re-
mains and those of site structure. While the former were advanc-
ing successful explanatory frameworks, the latter remained stagnant.
O’Connell suggested a simple, yet under-appreciated reason for this
difference: studies of faunal remains were building on the founda-
tions of a general theory which provided predictions a priori about
how individuals are expected to behave in particular circumstances,
but studies of site structure were based only on observations a pos-
teriori without any guiding theory. Because the former predictions
come from a general theory, they produce deductive inferences that
should be true in all cases and therefore do not need to rely on direct
⁎⁎ Corresponding author.
Email address: brian.codding@anthro.utah.edu (B.F. Codding)
ethnographic analogy in order to link ethnographic findings to the ar-
chaeological record. Because the later predictions come from empiri-
cal observations, they produce inductive inferences that must be tested
in every case and cannot escape the problems of direct ethnographic
analogy. It stands to reason then, for studies of site structure to move
forward, they must begin with a general theory. Yet despite this seem-
ingly simple fix, archaeologists guided by a general theory, such as
behavioral ecology, have failed to meet O’Connell’s challenge.
We argue that there is aclear and simple reason for this: because
movement at small spatial extents should not significantly constrain
behavior, patterning in site structure will either be explained by sim-
ple mechanical parameters (e.g., human body size; Binford, 1983) or
random variation. However, movement at larger spatial extents should
be significantly constrained by dynamic ecological patterning, and as
such, can be explained by a general theory such as evolution by natural
selection (e.g., Bird and O’Connell, 2006; Codding and Bird, 2015;
O’Connell, 1995). If this is true, then site structure can only be in-
formed by a general theory of behavior inasmuch as site-level pat-
terning can provide insight into larger, landscape scale decisions (in-
cluding subsistence strategies, settlement and mobility; e.g., Binford,
1980; Kent, 1991).
In an attempt to incorporate studies of site structure within a gen-
eral theory of behavior, here we link two common empirically-de-
rived, mechanistic predictions about site structure to two theoreti-
cally-derived, adaptive predictions from a simple behavioral ecolog-
ical model known as the Marginal Value Theorem (MVT, Charnov,
http://dx.doi.org/10.1016/j.jaa.2016.07.011
0278-4165/© 2016 Published by Elsevier Ltd.

UNCORRECTED PROOF
2 Journal of Anthropological Archaeology xxx (2016) xxx-xxx
1976; Charnov and Parker, 1995). Then we test these predictions with
ethnoarchaeological data collected in collaboration with Martu, an
Aboriginal population who reside in and have native title over their
ancestral estates in Western Australia.
2. Predictions
Behavioral ecological archaeologists focus on developing adaptive
explanations of past human decisions that can be tested with mate-
rial remains (Bird and O’Connell, 2006; Codding and Bird, 2015).
As outlined by Tinbergen (1963), adaptive explanations examine the
evolutionary function of behaviors in how they contribute to survival
and reproductive success. But this is only one of four levels of ex-
planations commonly employed to explain behavior. The others in-
clude mechanistic explanations, which focus on proximate causal fac-
tors, ontogenetic explanations, which focus on understanding how be-
haviors develop through the life course, and phylogenetic explana-
tions which focus on how a particular behavioral trait evolved within
a linage (Tinbergen, 1963). As discussed above, predictions about site
structure typically fall within the mechanistic level. Here we attempt
to link two of these mechanistic proposals to adaptive predictions de-
rived from the MVT.
2.1. Mechanistic predictions
Several mechanistic “rules”of site structure have been proposed
and tested with ethnographic or ethnoarchaeological data over the last
seventy-five years (reviews in David and Kramer, 2001). Two of these
are of particular relevance to the Marginal Value Theorem. These in-
volve site size and the degree of size sorting.
First, because multiple people cannot occupy the same space while
completing domestic tasks, the size of a site should increase with the
number of occupants, of course, with all else being equal (e.g., tasks
undertaken, kin-based residential rules, need for defense, etc.). This
pattern was first described and tested by Cook and Treganza (1950)
and subsequently confirmed by prominent ethnoarchaeologists work-
ing across the world (e.g., Yellen, 1977).
Second, because small discarded materials are likely to be dropped
in situ (e.g., Binford, 1978a) and because large discarded materials
will hinder ongoing activities within a site, people are likely to move
larger items away from central activity areas (Hayden and Cannon,
1983) while overlooking smaller items. This causes significant size
sorting wherein larger materials will tend to be more dispersed and
smaller materials will tend to remain near the point of primary de-
position (O’Connell, 1987). Importantly, the degree of size sorting
should be contingent on the duration of occupation: as individuals stay
longer at a site, it becomes worthwhile to remove obtrusive waste
(O’Connell, 1987, 100); at least up to some threshold where it be-
comes less costly to move to a new camp and start over (O’Connell,
1977). The amount of small material remaining in situ should de-
crease proportionally with the efficiency of the cleaning technology
(O’Connell, 1987, 92).
These predictions suggest that with hunter-gatherer camps, the size
of a site should be a reliable indicator of the number of people (and
number of active foragers if proportional to the number of people)
who occupied it and the degree of size sorting should provide infor-
mation on the duration of site occupation. Site size may also increase
with the actual (Yellen, 1977) or anticipated (e.g., Kent, 1992; Kent
and Vierich, 1989) duration of occupation, which may present a poten-
tial confound, but this is something that can be examined empirically.
2.2. Adaptive predictions
Based on the premise that natural selection should favor optimal
food acquisition strategies, the MVT provides a framework to under-
stand how long a forager should search for resources within a discrete
patch before leaving to travel to a neighboring patch (Charnov, 1976;
Charnov and Parker, 1995). Because a forager depresses the abun-
dance of resources by removing them (Charnov et al., 1976), the en-
ergy acquired within a patch diminishes as a function of the time spent
in the patch (Fig. 1). Following Charnov and Parker (1995), the cumu-
lative energy gain ( ) within a patch can be described as:
where tis the in-patch (foraging) time, is amount of energy orig-
inally in the patch (i.e., pre-foraging) and cis the rate at which ac-
quired energy reaches . Because the cumulative gain diminishes
over time, there should be an optimal threshold at which a forager
should leave the patch and travel to another patch. Charnov and Parker
(1995) approximate the optimal leave time ( ) as:
where is the travel distance to the next nearest patch. The deple-
tion rate (c) should be inversely proportional to the available energy in
the patch ( ) as a result of the increased time required to search the
patch (if prey density remains constant; Charnov and Parker, 1995).
In other words, larger patches have more resources, but take longer to
search.
This model provides a clear framework to predict how the number
of foragers within a patch and patch residence time should interact to
the determine the amount of energy acquired per capita. Holding the
available energy within a patch constant, when there are more people
within a patch, per capita gains will be lower (divided between more
Fig. 1. Graphical representation of the Marginal Value Theorem after Charnov and
Parker (1995). From a baseline (a; ), doubling the num-
ber of foragers in the patch should halve the energy acquired per capita, halve the
time required to search the patch and decrease the optimal patch residence time (b;
), while doubling the energy available in the patch should in-
crease the time required to search the patch (if prey density is constant) and increase the
optimal patch residence time (c; ). Note, travel time to the
next nearest patch is held constant ( ).
(1)
(2)

UNCORRECTED PROOF
Journal of Anthropological Archaeology xxx (2016) xxx-xxx 3
people) and less time will be spent searching within the patch. Hold-
ing the number of foragers constant, when a patch has more energy,
individuals will spend more time searching the patch, which will in-
crease the optimal patch residence time. Fig. 1 illustrates these predic-
tions with hypothetical scenarios.
While this model is specific to environmental situations where re-
sources are patchily distributed, it holds for humans even if patches
are homogeneous because humans are central place foragers (Orians
and Pearson, 1979) who return acquired food to a home base. As such,
human foraging segments environments into catchments around cen-
tral places.
2.3. Summary of predictions
If (P1) site size increases with the number of foragers and (P2) size
sorting increases with the duration of site occupation, then these attrib-
utes of site structure may be used as proxies for the number of foragers
within a patch and patch residence time, respectively. If so, then per
capita foraging returns should (P3) decrease with site size and (P4)in-
crease with size sorting. If these findings hold, these elements of site
structure may be incorporated within this general theory of behavior
to help explain variability in foraging behavior at larger spatial extents
where decisions are constrained by patterned environmental variation.
Following the strategy outlined in Codding and Bird (2015,
11–12), we test these four predictions using ethnographic and ethnoar-
chaeological data collected in collaboration with Martu.
3. Ethnographic background
Martu, also known as Mardu or Mardujarra, are part-time for-
agers who live on and have native title over their traditional estates
(sensu Stanner, 1965) in Western Australia. Within the climatolog-
ically-defined Arid Zone and the Western Desert cultural area, the
Martu homeland is centered on the Karlamilyi River, expanding west
to the Pilbara, northeast to the Percival Lakes and south to Lake Dis-
appointment (Fig. 2). The Native Title Determination Area includes
many of the estates belonging to Kartujarra, Manyjilyjarra, Warnman
dialect speakers, though the core Warnman estates around Karlamilyi
were retained by the Australian Government as a National Park.
Prior to European contact in the mid-twentieth century, Martu were
full-time foragers. Subsistence strategies varied by gender: men fo-
cused on larger hunted game, both men and women pursued smaller
game, and women focused on key plant resources (Tonkinson, 1993;
Veth and Walsh, 1988; Walsh, in press, 2008). Group structure was
centered on female kin who formed cooperative partnerships (Scelza
and Bliege Bird, 2008) and provided the majority of foraged foods
(Tonkinson, 1993, 43–45; see also Gould, 1969). When successful,
men’s hunting contributed significant proportions of meat to the diet,
but since success was rare, men’s production was highly variable
(Tonkinson, 1993, 43–45). Children foraged as well, sometimes help-
ing to gather fruits or hunt small game, with the proceeds of their labor
often consumed on the spot (Tonkinson, 1993, 47).
Many of these patterns continue today within a hybrid economy
(sensu Altman, 2001) wherein hunting and gathering for wild foods
remains one of the best economic alternatives (Codding et al., 2016).
Men tend to focus on larger prey which comes with a higher risk of
acquisition failure, while women focus on smaller, more reliable re-
sources (Bliege Bird and Bird, 2008; Bliege Bird et al., 2009; Codding
et al., 2011). Men’s hunting occasionally contributes significantly to
daily calories (Bliege Bird et al., 2009; Codding et al., 2011), but
given the high risk of failure, some aspects of men’s hunting may
be driven in part by traditional forms of social and ritual competi-
tion rather than food (Bird and Bliege Bird, 2010). Such competi-
tion may have real material outcomes, including lowering the age
of marriage for offspring (Scelza, 2010). Women’scooperative part-
nerships remain extremely important for both resource acquisition
(Bliege Bird et al., 2012b) and childcare (Scelza, 2009). The reli-
ability of women’s hunting is increased through the use of anthro-
pogenic fire, which feeds back to structure broad and varied envi-
ronmental patterns (Bird et al., 2005; Bliege Bird et al., 2008, 2012a,
2013; Codding et al., 2014; Zeanah et al., 2015). Children also remain
active foragers, sometimes contributing significant amounts of food to
Fig. 2. Areas surrounding Parnngurr within Martu Native Title area (inset).

UNCORRECTED PROOF
4 Journal of Anthropological Archaeology xxx (2016) xxx-xxx
their daily diet (Bird and Bliege Bird, 2005) and overall, foraging
remains an important source of provisioned foods for dependents
(Codding et al., 2016).
One major aspect of foraging that has changed centers on mobil-
ity. While traditionally charactered as high mobility foragers (sensu
Binford, 1980) who needed to move once local resources were dimin-
ished (Tonkinson, 1993), today Martu use vehicles to operate logis-
tically (sensu Binford, 1980) from a central community. This logis-
tic mobility facilitates a specific type of temporary settlement called a
dinner-time camps (see, e.g., Meehan, 1982).
3.1. Dinner-time camp foraging
A foraging party will leave the community and travel to a named
hunt region where they will establish a dinner-time camp. As the cen-
tral locus of contemporary foraging, these dinner-time camps are ex-
tremely important as a focal node of economic production and social
formation. While some dinner-time camps may be occupied repeat-
edly, most are used only once and produce very ephemeral material
signatures. Once foragers arrive at the camp, some may stay behind
to collect wood and build a fire, while others travel out in search of
wild resources. With most foraging activities, individuals travel out
from the dinner-time camp on foot and begin searching for target prey.
Two of the most common prey types are sand monitor lizards (also
known as goanna, Varanus gouldii, or parnajalpa) and hill kanga-
roo (also known as euro, Macropus robustus,orkirti-kirti). The for-
mer are found throughout the vast sand plains that blanket the area
while the latter are home in rocky ranges that dot the landscape (see
Fig. 2). Often these habitats occur side-by-side, allowing individual
hunters to walk out on foot from a centrally located dinner-time camp
and pursue either resource. While there is overlap in women’s and
men’sactivities (see above), often men walk out in search of hill kan-
garoo while women walk out onto the sand plain in search of monitor
lizards (Bliege Bird and Bird, 2008; Bliege Bird et al., 2009; Codding
et al., 2011). Because monitor lizards are a much more reliable re-
source (Codding et al., 2011), they provide the bulk of acquired calo-
ries returned to dinner-time camps (Bliege Bird and Bird, 2008).
We refer to the total time an individual spends searching for and
pursuing resources within a patch or foraging activity as aforaging
bout. Individuals rarely engage in more than one foraging bout per
day, making Martu dinner-time camp foraging not completely consis-
tent with the original formulation of the MVT, which was designed
to evaluate the optimal time a forager should spend in a patch before
leaving to travel to another patch. However, the model is still applic-
able in this case. Since foragers only undertake one bout per day, and
the decision to leave a patch is contingent on an individuals assess-
ment of opportunity costs, the general MVT framework still holds.
Once an individual decides their foraging bout is over, they return to
the dinner-time camp where everyone will process, cook and share
their foods based on traditional practices (e.g., Bird and Bliege Bird,
2010).
Unlike more permanent camps from pre-contact times (e.g., Gould,
1977), dinner-time camps rarely have any formal structures or features
beyond the roasting hearth. Hearths are the focus of activities at din-
ner-time camps. With smaller prey, like monitor lizards, the individual
hunter is generally responsible for processing and cooking their catch.
This involves removing the intestines, singeing the skin in hot flames,
digging an appropriate sized pit, lining the pit with hot coals, bury-
ing the prey in hot sand, and placing hot coals and burning wood on
top of the roasting pit. With larger prey, like hill kangaroo, it is the
responsibility of an available relative, preferably a mother’s brother
(e.g., Gould, 1967), to take over processing and cooking. The gen-
eral procedure is the same as with small prey, though with the dis
emboweling effort and the size roasting pit scaled proportionately to
the size of the animal (see also Bird and Bliege Bird, 2010; O’Connell
and Marshall, 1989). Once in the ground, cooking is passive and time
is generally passed with conversations about the hunt, the impending
meal, and life in general. Because they are only used once, hearths
themselves have relatively low fidelity: ethnoarchaeological investi-
gations reveal that while they are still identifiable eight years after cre-
ation, they are more subtle and diffuse (Codding, 2012); we expect
they would not be identifiable for more than one or two decades.
Once cooking is complete, prey are distributed among all who are
present (see Bird and Bliege Bird, 2010; Bird and Power, 2015; Bliege
Bird et al., 2012b; Codding, 2012, also Gould, 1967 for more details
on food sharing). Smaller prey are generally distributed whole while
larger prey are disarticulated according to traditional law (see Bird
and Bliege Bird, 2010). As individuals finish consuming their share,
some bones are dropped in place while others are gently tossed to the
side. In some ways this should produce material patterning similar to
Binford’s (1978a) “drop and toss”zones, which may result in size
sorting (O’Connell, 1987). Once everyone has had their fill, the for-
aging party packs up into the vehicles and returns to the community.
Most often all wild foods are consumed at the dinner-time camp, but
when harvests are abundant, remains may be taken back to the com-
munity for later consumption of further distribution.
4. Methods
4.1. Data collection
This study is part of a long-term ethnographic collaboration with
Martu that began in 2000. Intensive quantitative ethnographic data
used here were collected from 2000 to 2010 and include 1806 fo-
cal-individual follows (sensu Altmann, 1974) across 344 discrete din-
ner-time camps. The main unit of analysis is the individual foraging
bout: the total time spent on search and pursuit within an activity or
patch and the total energy acquired during that time. Greater detail on
data collection methods and on summary analyses of these data can be
found elsewhere (e.g., Bird et al., 2009; Bliege Bird and Bird, 2008;
Bliege Bird et al., 2009, 2012b; Codding et al., 2010, 2011, 2014).
Ethnoarchaeological investigations focused on a representative
sample of dinner-time camps where either sand monitor or hill kan-
garoo hunting took place. Of the 344 discrete dinner-time camps, in-
dividuals targeted either sand monitors or hill kangaroo at 196 camps
and twelve of these were selected for ethnoarchaeological investiga-
tions. Quantitative foraging data from this ethnoarchaeological subset
do not differ significantly from the main foraging dataset, indicating
that it is a representative sample (Codding, 2012). Ethnoarchaeologi-
cal excavations were undertaken between 2009 and 2010. Researchers
revisited the selected dinner-time camps and established a grid cen-
tered on the hearth dividing the entire site into 1-x-1-m units. Soils
were excavated with shovel scrapes within each 1-x-1-m unit that con-
tinued until sterile sand (typically < 5 cm). Materials were screened
through nested 6-mm (1/4-in.) and 3-mm (1/8-in.) mesh, providing a
large and a small fraction that could be used to examine variation
in size sorting across the site. Analyses focused on faunal material.
Greater detail on these methods and additional analyses are reported
in Codding (2012).
From these combined datasets, we extract six variables to test the
four main predictions listed above: site size, the number of site occu-
pants (and how many of those are foragers), the degree of size sort-
ing, the duration of site occupation and both the median and mean per
capita monitor lizard harvest. Site size is calculated as the number of

UNCORRECTED PROOF
Journal of Anthropological Archaeology xxx (2016) xxx-xxx 5
1-x-1-m units from which bones were recovered. The number of site
occupants and the number of foragers are tallied from quantitative ob-
servations. Size sorting is calculated as the standard distance devia-
tion (see below) of the 6-mm fraction relative to the 3-mm fraction.
The duration of active site occupation (or occupation time) is the total
person minutes at the site excluding time spent foraging. This should
measure the cumulative impact of human activity at the site, which
we expect will lead to greater size sorting. Because the distribution of
monitor harvest (kcal per bout) is skewed (see Codding et al., 2010),
mean and median values per camp may provide a significantly differ-
ent result. As such, we run analyses using both mean and median per
capita harvest size (kcals) per bout.
4.2. Data analysis
All analyses are performed in the R environment (R Core Team,
2015). Here we review our approach to estimating size sorting and to
hypothesis testing.
The dispersion of bone fragments was quantified for each site as
the standard distance deviation (or standard distance). This was im-
plemented in the aspace library (Bui et al., 2012) using the number
of bones recovered from each 1-x-1-m unit for each recovery method
(fine and coarse grained mesh). For visualization purposes, standard
deviation ellipses were also calculated for each recovery method. Size
sorting was then calculated as the difference in standard distance be-
tween the 3-mm and 6-mm factions so that the more positive numbers
equal greater dispersion of larger bone fragments relative to smaller
fragments.
Tests of predictions rely on linear models. However, ordinary least
squares regression models are inappropriate for this analysis because
the independent (response) variables are either heavily skewed (i.e.,
not normally distributed) or not continuous (e.g., count data). In-
stead of turning to rank-ordered non-parametric statistics, which re-
duce variation by converting continuous or count data to ordinal num-
bers, we rely on generalized linear models (GLM) designed to extend
Gaussian linear models by specifying a distribution and link function
appropriate to the data (Faraway, 2006; R Core Team, 2015). Unless
otherwise noted in the text, we rely on models with a Poisson distribu-
tion, log link and quasi-likelihood estimation. Poisson-log GLMs with
quasi-likelihood estimation are appropriate for our data which are ei-
ther count (e.g., number of occupants) or count-like (e.g., number of
calories), and non-normally distributed.1Through quasi-likelihood es-
timation, these models also avoid over-dispersion by relaxing the as-
sumption that the mean and variance are equal. As with GLMs gen-
erally, quasi-likelihood Poisson models are common in ecology (e.g,
Ohara and Kotze, 2010) and their application is growing in archae-
ology (e.g., Shennan et al., 2013). Model results report the Y-inter-
cept, estimated change in the dependent (y) variable with a change in
the independent (x) variable, or p-value associated with the inde-
pendent variable (Ind. p), log-likelihood r-square value which
is the proportion of deviance explained by the inclusion of the inde-
pendent (predictor) variable, and the p-value from a likelihood ratio
1Results of Shapiro–Wilk Normality Test reveal that the dependent (response)
variables are not normally distributed: number of occupants
, the number of foragers
, occupation time ,
foraging time , median harvest
, and mean harvest
. While the median sand monitor harvest only departs from normal with marginal
significance ( ), it is sampled from a larger distribution of sand monitor
hunts that is highly skewed .
test (LRT p) which indicates whether or not the inclusion of the
predictor variable significantly improves the model fit compared to
the null model. Because these models incorporate a log-link, y-inter-
cepts and estimates are reported as the exponent for interpretation.
5. Results
Before testing the specific predictions, analyses must confirm that
monitor lizard foraging follows the general framework proposed by
the marginal value theorem and confirm that size sorting is present at
these sites.
As shown in Fig. 3, across 166 camps at which at least one individ-
ual went sand monitor hunting, the mean per capita harvest (E, kcal)
varies significantly as a function of the time spent in patch (T, min)
when fit with a normal-log linear model .
This can be approximated by a diminishing returns curve following
the MVT . This is also true for the median
per capita harvest .
These findings suggest that monitor lizard hunting, and the represen-
tative sample of ethnoarchaeological camps, should provide a viable
window into MVT dynamics .
Size sorting was calculated from bone counts per unit, which
ranged from zero to 33 in the 6-mm fraction and from zero to 43 in
the 3-mm fraction (Fig. 4). Data on bone dispersion is available for
ten of the twelve sites and monitor lizard hunting occurred at nine of
these. The two sites for which dispersion data is unavailable include
one where only a failed kangaroo hunt occurred, producing no mater-
ial signature other than a hearth, and one where only a single bone was
recovered, from which it is impossible to calculate dispersion. Of the
remaining ten sites, larger bone fragments recovered from the coarse
grained (6-mm) fraction are generally more dispersed than smaller
bone fragments recovered from the fine grained (3-mm) fraction. In-
cluding all ten sites, the difference in dispersion between ranges from
−0.85 to 0.41, with non-negative value for the mass of the distribu-
tion (7 out of 10 sites; IQR = −0.12–0.22, median = 0.032). Exclud-
ing the one site where monitor lizard hunting did not occur, disper-
sion across the nine remaining sites confirms that larger material tends
to be more dispersed than smaller material: the difference in standard
distance ranges from −0.27 to 0.41
Fig. 3. Mean per capita monitor lizard harvest (E, kcal) as a function of the time spent in
patch (T, min) across 166 sand monitor hunting camps shown with a normal-log GLM
fit (black solid line with dashed 95% confidence intervals) and an approximate dimin-
ishing returns curve following the MVT (gray solid line). Ethnoarchaeological camps
shown as solid black points.

UNCORRECTED PROOF
6 Journal of Anthropological Archaeology xxx (2016) xxx-xxx
Fig. 4. Distribution of recovered bones from 3-mm mesh (left) and 6-mm mesh (right) across 1-x-1-m units superimposed for all 11 excavated ethnoarchaeological camps. Relative
density of recovered bones per unit illustrated in gray-scale from none in white to maximum (43 and 33 respectively) in black. Standard deviation ellipses (circles) indicate that larger
bone fragments are more dispersed overall. The center of dispersion (points) for both recovery methods remains close to hearth locations (located at 5.5 m North, 5.5 m East), with
the 3-mm fraction being slightly closer.
(IQR = 0.00–0.26, median = 0.06) with only two sites exhibiting neg-
ative size sorting. These findings illustrate that significant size sort-
ing generally occurs even at these ephemeral, short-term dinner-time
camps.
These results confirm that the foraging data examined here does
conform to the MVT assumptions, and that size sorting is occurring at
these sampled camps. This allows analyses to move forward with the
examination of each of the mechanistic and adaptive predictions pro-
posed above.
5.1. Prediction 1: the number of foragers should increase with site
size
Based on the simple mechanics of how humans occupy space, the
number of site occupants and (if proportional) the number of foragers
should increase as a function of site size. Across the eleven excavated
sites, the number of site occupants and the number of active foragers
significantly increase with site size (Table 1, Fig. 5a). This confirms
the first mechanistic prediction and suggests that site size may be a
useful proxy of foraging yields.
5.2. Prediction 2: the duration of occupation should increase with
size sorting
If material becomes more dispersed as a function of more time
spent a the site, then the duration of occupation should increase with
degree of size sorting. Across the ten sites for which size sorting data
Table 1
Summary of bivariate model results testing each of the four predictions.
Pred. Dep. var. Ind. var. Y-intcp. Est. Ind. pLRT p∗
1 # of
Occupants
Site size 7.05 1.02 0.0696 0.31 0.0314
1 # of Foragers Site size 2.71 1.03 0.0517 0.35 0.0111
2 Occupation
time
Size
sorting
2624.91 3.29 0.3310 0.16 <0.0001
2 Foraging time Size
sorting
384.87 3.88 0.3690 0.12 <0.0001
3 Median
harvest
Site size 2744.10 −1.02 0.4180 0.10 <0.0001
3 Mean harvest Site size 2419.22 −1.00 0.9090 <0.01 <0.0001
4 Median
harvest
Size
sorting
1563.53 6.83 0.1060 0.28 <0.0001
4 Mean harvest Size
sorting
1879.76 5.34 0.0768 0.38 <0.0001
∗∗ Note: LRT p-values report the results of a likelihood ratio test, which indicates if the
inclusion of the independent variable significantly improves the model fit compared to
a null model.
was available, the active occupation time and the sum time spent hunt-
ing monitor lizards increases as a function of the difference in disper-
sion between large and small bone fragments (Fig. 5b). While overall
the models are a significant improvement over a null model, the ef-
fect of each independent variable is not statistically significant (Table
1). These findings suggest that size sorting may be a rough proxy for
the amount of time individuals spend at a site and the amount of time
spent foraging, but that the effect may sometimes be due to chance.
5.3. Prediction 3: overall foraging yields should decrease with site
size
Based on theoretical insights from the MVT, more people, occupy-
ing larger sites, should cause greater in-patch competition and result
in lower per capita yields. The results show that this generally holds
true: monitor lizard harvests decline as site size increases (Fig. 5c).
But the effect is only significant inasmuch as the inclusion of site size
improves the prediction of monitor lizard returns when compared to
a null model ( ), with any independent effects of site size
likely due to chance (Table 1). Moreover, this is not a very robust as-
sociation, with site size explaining only 10% of deviance in median
returns and less than 1% in mean returns.
5.4. Prediction 4: overall foraging yields should increase with size
sorting
Foragers should stay longer in higher quality patches, which
should result in greater size sorting and higher per capita returns. The
data support this prediction, showing that mean and median harvests
increase with the degree of size sorting (Fig. 5d). Size sorting explains
28% of the deviance in median monitor lizard harvest and 38% in
mean harvest. Though while the inclusion of the independent variable
improves each model significantly , the effect of each in-
dependent variable is only marginally significant ( ; Table 1).
Overall, these findings roughly support the predictions, showing
that per capita foraging yields decrease with site size and increase with
size sorting. However, while the results do move in the predicted di-
rections, the effects are less robust than would be ideal. These limited
effects may result from the interactions between the number of occu-
pants and patch duration.

UNCORRECTED PROOF
Journal of Anthropological Archaeology xxx (2016) xxx-xxx 7
Fig. 5. Plots and model fits summarizing results from the four main predictions. (a, P1) Number of site occupants as a function of site size ( ). (b, P2) Duration of active site
occupation as a function of size sorting. (c, P3) Median monitor lizard harvest (kcals) as a function of site size and (d, P4) as a function of size sorting. Confidence intervals show the
standard error (95%) of the model fit. See Table 1 for details.
5.5. Interactive effects
To deal with the potential interactive effects of the number of for-
agers and the duration of occupation (and their material correlates)
on overall returns, we analyze these in a series of multivariate mod-
els. Combined, the number of foragers and the duration of occupation
accurately predict median and mean monitor lizard harvest (Table 2,
Fig. 6). This is especially the case with median harvests, where each
independent variable is marginally significant and the overall model
is highly significant explaining 58% of the deviance. These patterns
are consistent into the material record: site size and the degree of size
sorting also accurately predict monitor lizard harvest (Table 2, Fig. 6).
Again, these patterns are more robust with median returns, of which
site size and size sorting have a marginally significant and significant
effect respectively, together explaining 50% of the deviance. In both
cases, these findings are more robust with median than mean returns
and indicate that size sorting might be a more significant predictor of
overall returns than site size.
6. Discussion
At a descriptive level, these findings confirm previous results il-
lustrating that site size tends to increase with the number of occupants
(e.g., Yellen, 1977) and that smaller artifacts are more likely to remain
in their primary context than are larger objects (e.g., O’Connell, 1987).
The later finding is quite remarkable given that these are ephemeral,
short-term camps but size sorting is still present in seven of the nine
focal sites.
Table 2
Summary of multivariate model results examining the combined effect of each indepen-
dent variable on mean and median monitor lizard harvest size (kcal) per bout.
Dep. var. Ind. var. Y-intcp. Est. Ind. pLRT ∗
Median
harvest
Whole model 3418.65 – – 0.58 <0.0001
# Foragers –−0.80 0.0829 – –
Occupation
time
–1.00 0.0930 – –
Mean harvest Whole model 2342.56 – – 0.25 <0.0001
# Foragers –−0.94 0.4950 – –
Occupation
time
–1.00 0.2420 – –
Median
harvest
Whole model 2245.96 – – 0.50 <0.0001
Site size –−1.04 0.1059 – –
Size sorting –11.41 0.0399 – –
Mean harvest Whole model 2187.05 – – 0.44 <0.0001
Site size –−1.01 0.4383 – –
Size sorting –6.73 0.0715 – –
∗∗ Note: LRT p-values report the results of a likelihood ratio test, which indicates if the
inclusion of the independent variables significantly improves the model fit compared
to a null model.

UNCORRECTED PROOF
8 Journal of Anthropological Archaeology xxx (2016) xxx-xxx
Fig. 6. Plots summarizing two multivariate models: the first (top row) examining the combined effect of (1a) the number of foragers and (1b) the duration of site occupation (min) on
median per capita monitor lizard harvest (E, kcals); the second (bottom row) examining the combined effect of (2a) site size ( ) and (2b) size sorting on median per capita monitor
lizard harvest (E, kcals). Model fits illustrate the effect of each independent variable while holding the other constant at its median value. Confidence intervals show the standard error
(95%) of the model fit. See Table 2 for details.
At an explanatory level, these findings illustrate that aspects of site
structure may be useful proxies of larger-scale ecological interactions
that lead to variation in foraging yields. Where foraging follows the
assumptions of the MVT, the combined interaction of site size and size
sorting should be indicative of overall wild food harvests. Although
we do not want to over-interpret our empirical findings. Our bivari-
ate model results are less robust than would be ideal, with relatively
high p-values suggesting that some of these results may be false pos-
itives (Type I errors). Nonetheless, all of the trends do move in the
predicted direction, which is encouraging in itself given the relatively
small size of the ethnoarchaeological dataset and all of the noise ex-
pected in these parameters.
The limitations of these bivariate models may also result from the
interactive effects of site size (as a proxy of foraging competition) and
size sorting (as a proxy of patch quality). When combined in a mul-
tivariate model to control for these interactions, the results show that
site size and size sorting can predict up to 50% of the noise in median
harvests, with less than a 10% chance that any one variable is produc-
ing a false positive. This result confirms the main predictions proposed
in this paper and highlight how these two measures of site structure
can be used to inform researchers on broader scale patterns in foraging
ecology.
In addition to supporting the major predictions, the results from
multivariate models suggest that these measures may be more indica-
tive of median than mean returns. Since median values are more ro-
bust to skew, this should be true with any resource where returns are
characterized by a skewed distribution biased by zeros resulting from
failed bouts. These findings also suggest that size sorting may be a
clearer indicator of overall foraging yields driven by patch quality than
site size may be of in-patch competition driven by the number of for-
agers. This could result because site size is more a reflection of the
number of occupants than the number of foragers, or it could be that
the number of foragers within a patch increases cooperation rather
than competition, which may lead to greater per capita yields (e.g.,
Smith, 1991, cf. Bliege Bird et al., 2012b).
6.1. Applications in archaeological contexts
This framework should significantly expand the explanatory role
of site structure in archaeological studies of hunter-gatherers. How-
ever, these findings will be applicable in archaeological contexts only
where at least two conditions are met, one relating to site formation
and the other to investigation. First, accurate patterns of material dis-
persion will only be preserved in rapid depositional environments that
are impacted by limited post-depositional activity. While this may
be rare (e.g., Fanning and Holdaway, 2001), it is not unknown (e.g.,
Enloe, 2006). Second, adequate reconstructions of site structure will
require investigations that excavate large, contiguous exposures and
carefully record material to capture intra-site variation in their distri-
bution (O’Connell, 1987, 1995). While previous work has cautioned
about the extraordinary costs associated with fine-grained analyses
of microrefuse (Metcalfe and Heath, 1990), here we show that these
interpretations are possible even with relatively coarse-grained exca-
vation techniques relying on a 1-x-1-m grid and nested 6-mm and
3-mm mesh screens. One final condition to consider concerns whether
the activities under investigation were centered around

UNCORRECTED PROOF
Journal of Anthropological Archaeology xxx (2016) xxx-xxx 9
hearths. Hearths provide a focal area for site occupants to center ac-
tivities (e.g., Binford, 1978b,a) and a durable indicator for archaeol-
ogists to center excavations (e.g., Metcalfe and Heath, 1990; Simms
and Heath, 1990); as such, the applicability of these insights to archae-
ological contexts will likely be amplified when hearth-centered activ-
ities are under investigation.
Where these criteria are met, including in archaeological contexts
in Africa (e.g., Parkington et al., 2009), the Near East (e.g., Maher et
al., 2012), Europe (e.g., Enloe, 2006), Asia (e.g., Sakaguchi, 2007),
and the Americas (e.g., Bamforth et al., 2005; Burns, 2005; Simms,
1989; Tipps, 1993), examinations of the relative differences in site
size and size sorting through time or across space may be used to
inform on variation in foraging efficiency and its consequences for
subsistence, settlement and mobility. For example, if the suite of re-
sources represented at a site and its size remain constant, a diachronic
shift toward greater size sorting may indicate higher environmental
productivity, producing higher foraging yields and lower mobility.
Or, if the suite of resources represented at a site widens to include
lower profitability resources, shifts toward larger sites and/or greater
size sorting could be used as supporting evidence of resource inten-
sification (i.e., Boserup, 1965, see Morgan, 2015), with more people
spending more time in patch taking lower profitability resources. Of
course, in any of these circumstances, multiple lines of evidence will
be needed to support these findings. Additional elements of site struc-
ture, such as the presence or absence of storage features, can also help
inform on the duration of occupation (Binford, 1980).
6.2. Future ethnoarchaeological work
While the findings presented here are encouraging, future ethnoar-
chaeological investigations are needed to confirm these results and ex-
amine how they vary in different contexts. Existing data from exten-
sive studies in Africa (e.g., Yellen, 1977; O’Connell et al., 1991) and
Australia (e.g., Hayden, 1979; O’Connell, 1987) could also be used
to evaluate some or all of these hypotheses. Additional questions may
also be approached within this general framework.
For example, variation in size sorting at longer-term camps may
be a function of the ways in which technologies alter cleaning effi-
ciency (Metcalfe and Heath, 1990; O’Connell, 1987). The costs and
benefits of different cleaning methods and technologies may present
a trade-off that can be modeled by the MVT: with increased duration
of occupation, investments in better cleaning technology might be off-
set by increased cleaning efficiency, which should lead to exponen-
tial increases in size sorting. This could be tested by examining con-
texts where hunter-gatherer settlement patterns produce some sites oc-
cupied for short intervals and others occupied for long intervals, both
with attendant differences in cleaning technology.
Future work could also use this framework to examine how pat-
terns in site structure vary by gender to reveal differences in the di-
vision of foraging labor. Specifically, where women’s and men’s for-
aging strategies produce different sites that differ consistently in the
number of foragers or in the time spent foraging, comparisons between
sites may reveal patterns in site size and size sorting that can be used to
differentiate women’s and men’s camps (e.g., O’Connell et al., 1991).
The true value of any of this work will be its ability to expand
our understanding of site structure from the perspective of a gen-
eral theory of behavior (O’Connell, 1995). Until additional ethnoar-
chaeological work is undertaken to validate and expand theory-driven
analyses of site structure, “…archaeologists interested in site struc-
ture are stuck with highly speculative predictions and interpretations
grounded in some combination of local ethnography, exotic ethnoar-
chaeology, and their own intuition”(O’Connell, 1993, 24).
7. Conclusion
Studies of site structure have remained largely descriptive, even
over the past twenty years since O’Connell (1995) illustrated the po-
tential utility of approaching these problems with a general theory of
behavior. Here we suggest this is the result of an inherent limitation
of site structure given that the behaviors which produce it are not sig-
nificantly structured by adaptive constraints at the site level. As an al-
ternative, we propose and implement an approach that uses measures
of sites structure as proxies for larger scale patterns which should be
structured by environmental variability. This is not new in itself (e.g.,
Binford, 1980), but is novel in linking these proxies to a general the-
ory of behavior that provides deductive, a priori predictions that avoid
the problems associated with direct ethnographic analogy (O’Connell,
1993, 1995). This approach does not completely resolve all the issues
of site structure—especially those resulting from formation or investi-
gation bias—but we hope this application will inspire future studies of
site structure to examine these and develop other avenues to link pat-
terned spatial variation in deposited material to larger ecological pat-
terns of human decisions that can be explained by a general theory of
behavior.
Acknowledgements
We owe a tremendous debt of gratitude to our Martu collabora-
tors for their guidance, friendship and support. Curtis Taylor provided
instrumental assistance in the implementation of this work. Special
thanks to Hamza, Chili, Wilson, Roderick, Nyaparu and Nyaparu for
the assistance in collecting these data. This paper was originally de-
veloped as a presentation in a session honoring James F. O’Connell
organized by Karen Lupo at the 80th annual meeting of the Society
for American Archaeology in San Francisco: thanks to Karen for in-
cluding us in such a successful event, to all the participants for help-
ful comments, and to Jim for the inspiration and lunch. Addition-
ally, thanks to Jim for recognizing the utility of a general theory of
behavior and for supporting all of our efforts to match his elegance
in their application to ethnographic and archaeological problems. We
are grateful for the support of our colleagues, including Peter Kauha-
nen, Brooke Scelza, Bob and Myrna Tonkinson, Peter Veth, Fiona
Walsh, and especially Sarah Robinson who provided a sounding board
for many of these incipient ideas. This paper benefited significantly
from comments by Duncan Metcalfe, Kenneth Blake Vernon, Kate
Magargal, and two anonymous reviewers. Financial support for this
work comes from the National Science Foundation (BCS-0314406,
BCS-0850664, DDIG BCS-0915380, BCS-1459880), the Woods In-
stitute for the Environment, Department of Anthropology and Archae-
ology Center at Stanford University.
References
Altman, J., 2001. Sustainable Development Options on Aboriginal Land: The Hybrid
Economy in the Twenty-first Century. Center for Aboriginal Economic Policy Re-
search Discussion Paper, vol. 226. pp. 1–13.
Altmann, J., 1974. Observational study of behavior: sampling methods. Behav-
iour 91, 449–459.
Bamforth, D.B., Becker, M., Hudson, J., 2005. Intrasite spatial analysis, ethnoarchaeol-
ogy, and paleoindian land-use on the great plains: the allen site. Am. An-
tiq. 70, 561–580.
Binford, L., 1983. In: Pursuit of the Past: Decoding the Archaeological Record.
Thames and Hudson, New York.

UNCORRECTED PROOF
10 Journal of Anthropological Archaeology xxx (2016) xxx-xxx
Binford, L.R., 1978. Dimensional analysis of behavior and site structure: learning from
an Eskimo hunting stand. Am. Antiq. 43, 330–361.
Binford, L.R., 1978. Nunamuit Ethnoarchaeology, Academic Press, New York.
Binford, L.R., 1980. Willow smoke and dogs’tails: hunter-gatherer settlement systems
and archaeological site formation. Am. Anthropol. 45, 4–20.
Bird, D.W., Bird, R.B., Parker, C.H., 2005. Aboriginal burning regimes and hunting
strategies in Australia’s Western Desert. Hum. Ecol. 33 (4), 443–464.
Bird, D.W., Bliege Bird, R., 2005. Martu children’s hunting strategies in the Western
Desert, Australia. In: Hewlett, B.,Lamb, M. (Eds.), Hunter-Gatherer Childhoods:
Evolutionary, Developmental & Cultural Perspectives. Aldine Transactions, New
Brunswick, pp. 129–146.
Bird, D.W., Bliege Bird, R., 2010. Competing to be leaderless: food sharing and mag-
nanimity among Martu Aborigines. In: Kantner, J., Vaughn, K. (Eds.), The Emer-
gence of Leadership: Transitions in Decision Making from Small-Scale to Mid-
dle-Range Societies. School of American Research, Santa Fe.
Bird, D.W., Bliege Bird, R., Codding, B.F., 2009. In pursuit of mobile prey: Martu
hunting strategies and archaeofaunal interpretation. Am. Antiq. 74 (1), 3–29.
Bird, D.W., O’Connell, J.F., 2006. Behavioral ecology and archaeology. J. Archaeol.
Res. 14, 143–188. http://dx.doi.org/10.1007/s10814-006-9003-6.
Bird, R.B., Power, E.A., 2015. Prosocial signaling and cooperation among Martu
hunters. Evol. Hum. Behav. 36 (5), 389–397.
Bliege Bird, R., Bird, D.W., 2008. Why women hunt: risk and contemporary foraging
in a Western Desert Aboriginal community. Curr. Anthropol. 49, 655–693.
Bliege Bird, R., Bird, D.W., Codding, B.F., Parker, C.H., Jones, J.H., 2008. The “fire
stick farming”hypothesis: Australian Aboriginal foraging strategies, biodiversity,
and anthropogenic fire mosaics. Proc. Nat. Acad. Sci. 105 (39), 14796–14801.
Bliege Bird, R., Codding, B.F., Bird, D.W., 2009. What explains differences in men’s
and women’s production? determinants of gendered foraging inequalities among
Martu. Hum. Nat. 20, 105–129. http://dx.doi.org/10.1007/s12110-009-9061-9.
Bliege Bird, R., Codding, B.F., Kauhanen, P.G., Bird, D.W., 2012. Aboriginal hunting
buffers climate-driven fire-size variability in Australia’s spinifex grasslands. Proc.
Nat. Acad. Sci. 109, 10287–10292.
Bliege Bird, R., Scelza, B., Bird, D.W., Smith, E.A., 2012. The hierarchy of virtue:
mutualism, altruism and signaling in Martu women’scooperative hunting. Evol.
Hum. Behav. 33, 64–78.
Bliege Bird, R.,Taylor, N., Codding, B.F., Bird, D.W., 2013. Niche construction and
Dreaming logic: Aboriginal patch mosaic burning and varanid lizards (Varanus
gouldii) in Australia. Proc. Roy. Soc. B 280, 20132297.
Boserup, E., 1965. The Conditions of Agricultural Growth: The Economics of Agrar-
ian Change under Population Pressure. Aldine, Chicago, Illinois.
Broughton, J.M., Cannon, M.D., Bayham, F.E., Byers, D.A., 2011. Prey body size and
ranking in zooarchaeology: theory, empirical evidence and applications from the
northern Great Basin. Am. Antiq. 76, 403–428.
Bui, R., Buliung, R.N., Remmel, T.K., 2012. Aspace: a collection of functions for esti-
mating centrographic statistics and computational geometries for spatial point pat-
terns. v. 3.2.
Burns, J.A., 2005. What about behavior? Methodological implications for rockshelter
excavation and spatial analysis. North Am. Archaeol. 26 (3), 267–282.
Byrd, B.F., Garrard, A.N., Brandy, P., 2015. Modeling foraging ranges and spatial or-
ganization of Late Pleistocene hunter-gatherers in the southern Levant: a least-cost
GIS approach. Quatern. Int.
Charnov, E., Parker, G.A., 1995. Dimensionless invariants from foraging theory’smar-
ginal value theorem. Proc. Nat. Acad. Sci. 92, 1446–1450.
Charnov, E.L., 1976. Optimal foraging, the marginal value theorem. Theor. Popul.
Biol. 9, 129–136.
Charnov, E.L., Orians, G., Hyatt, K., 1976. Ecological implications of resource depres-
sion. Am. Nat. 110, 247–259.
Codding, B.F., 2012. ‘Any Kangaroo?’On the Ecology, Ethnography and Archaeology
of Foraging in Australia’s Arid West. PhD thesis, Department of Anthropology,
Stanford University.
Codding, B.F., Bird, D.W., 2015. Behavioral ecology and the future of archaeological
science. J. Archaeol. Sci. 56, 9–20.
Codding, B.F., Bird, D.W., Bliege Bird, R., 2010. Interpreting abundance indices:
some zooarchaeological implications of Martu foraging. J. Archaeol. Sci. 37 (12),
3200–3210.
Codding, B.F., Bliege Bird, R., Bird, D.W., 2011. Provisioning offspring and others:
risk-energy trade-offs and gender differences in hunter-gatherer foraging strate-
gies. Proc. Roy. Soc. B 278, 2502–2509.
Codding, B.F., Bliege Bird, R., Bird, D.W., Zeanah, D.W., 2016. Alternative Aborigi-
nal economies: Martu livelihoods in the 21st century. In: Codding, B.F., Kramer,
K.L. (Eds.), Why Forage? Hunters and Gatherers in the 21st Century. School for
Advanced Research Press, Santa Fe and University of New Mexico Press, Albu-
querque.
Codding, B.F., Bliege Bird, R., Kauhanen, P.G., Bird, D.W., 2014. Conservation or
co-evolution? Intermediate levels of Aboriginal hunting and burning have positive
effects on kangaroo populations in Western Australia. Hum. Ecol. 42, 659–669.
http://dx.doi.org/10.1007/s10745-014-9682-4.
Cook, S.F., Treganza, A.E., 1950. The quantitative investigation of Indian mounds:
with special reference to the relation of the physical components to the probable
material culture. Univ. Calif. Publ. Am. Archaeol. Ethnol. 40, 223–262.
David, N., Kramer, C., 2001. Ethnoarchaeology in Action. Cambridge University
Press.
Enloe, J.G., 2006. Geological processes and site structure: assessing integrity at aLate
Paleolithic open-air site in northern France. Geoarchaeology 21 (6), 523–540.
Fanning, P., Holdaway, S., 2001. Stone artifact scatters in western NSW, Australia: ge-
omorphic controls on artifact size and distribution. Geoarchaeology 16 (6),
667–686.
Faraway, J., 2006. Extending the Linear Model with R: Generalized Linear, Mixed Ef-
fects and Nonparametric Regression Models. Chapman and Hall, New York.
Gould, R.A., 1967. Notes on the hunting, butchering, and sharing of game among the
Ngatatjara and their neighbors in the West Australian desert. Kroeber Anthropol.
Pap. 36, 41–66.
Gould, R.A., 1969. Subsistence behaviour among the Western Desert Aborigines of
Australia. Oceana 4, 253–274.
Gould, R.A., 1977. Puntutjarpa Rockshelter and the Australian Desert Culture. Anthro-
pological Papers, vol. 54. The American Museum of Natural History.
Hayden, B., 1979. Palaeolithic Reflections: Lithic Technology and Ethnographic Exca-
vation among Australian Aborigines. Australian Institute of Aboriginal Studies,
Canberra.
Hayden, B.,Cannon, A., 1983. Where the garbage goes: refuse disposal in the Maya
Highlands. J. Anthropol. Archaeol. 2, 117–163.
Hill, M.G., Rapson, D.J., Loebel, T.J., May, D.W., 2011. Site structure and activity or-
ganization at aLate Paleoindian base camp in Western Nebraska. Am. An-
tiq. 76 (4), 752–772.
Kelly, R.L., Surovell, T.A., Shuman, B.N., Smith, G.M., 2013. A continuous climatic
impact on Holocene human population in the Rocky Mountains. Proc. Nat. Acad.
Sci. 110 (2), 443–447.
Kent, S., 1991. The relationship between mobility strategies and site structure. In:
Kroll, E., Price, T.D. (Eds.), The Interpretation of Spatial Patterning Within Stone
Age Archaeological Sites. Plenum Press, New York, pp. 33–59.
Kent, S., 1992. Studying variability in the archaeological record: an ethnoarchaeologi-
cal model for distinguishing mobility patterns. Am. Antiq. 57, 635–660.
Kent, S., Vierich, H., 1989. The myth of ecological determinism-anticipated mobility
and site spatial organization. In: Kent, S. (Ed.), Farmers as Hunters-Implications of
Sedentism. Cambridge University Press, pp. 96–134.
Maher, L., Richter, T., Stock, J.T., MacDonald, D., Jones, M., Martin,L., 2012.
Twenty thousand-year-old huts at a hunter-gatherer settlement in eastern Jordan.
PLoS One 7 (2), 1–10.
Meehan, B., 1982. Shell Bed to Shell Midden. Australian Institute of Aboriginal Stud-
ies,Canberra.
Metcalfe, D., Heath, K.M., 1990. Microrefuse and site structure: the hearths and floors
of the Heartbreak Hotel. Am. Antiq. 55, 781–796.
Morgan, C., 2012. Modeling modes of hunter-gatherer food storage. Am. An-
tiq. 77, 714–736.
Morgan, C., 2015. Is it intensification yet? Current archaeological perspectives on the
evolution of hunter-gatherer economies. J. Archaeol. Res. http://dx.doi.org/10.
1007/s10814-014-9079-3:1-51.
O’Connell, J., Hawkes, K., Jones, N., 1991. Distribution of refuse-producing activities
at Hadza residential base camps: implications for analyses of archaeological site
structure. In: Kroll, E., Price, T. (Eds.), The Interpretation of Archaeological Spa-
tial Patterning. Plenum, New York, pp. 61–76.
O’Connell, J.F., 1977. Room to move: contemporary Alyawara settlement patterns and
their implications for Aboriginal housing policy. Mankind 11, 119–131.
O’Connell, J.F., 1987. Alyawara site structure and its archaeological implications. Am.
Antiq. 52, 74–108.
O’Connell, J.F., 1993. What can Great Basin archaeologists learn from the study of site
structure? An ethnoarchaeological perspective. Utah Archaeol. 6, 7–26.
O’Connell, J.F., 1995. Ethnoarchaeology needs a general theory of behavior. J. Ar-
chaeol. Res. 3, 205–255. http://dx.doi.org/10.1007/BF02231450.
O’Connell, J.F., Marshall, B., 1989. Analysis of kangaroo body part transport among
the Alyawara of central Australia. J. Archaeol. Sci. 16 (4), 393–405.
Ohara, R.B., Kotze, D.J., 2010. Do not log-transform count data. Methods Ecol.
Evol. 1 (2), 118–122.
Orians, G.H., Pearson, N.E., 1979. On the theory of central place foraging. In: David,
J., Horn, G.R.S., Mitchell, R.T. (Eds.), Analysis of Ecological Systems. Ohio State
University Press, pp. 155–177.
Parkington, J., Fisher Jr., J.W., Tonner, T.W., 2009. “The fires are constant, the shel-
ters are whims”: a feature map of later stone age campsites at the Dunefield Mid-
den Site, Western Cape Province, South Africa. S. Afr. Archaeol.
Bull. 64, 104–121.
R Core Team, 2015. R: ALanguage and Environment for Statistical Computing. R
Foundation for Statistical Computing, Vienna, Austria. v.3.2.3 (2015-12-10,
Wooden Christmas-Tree).
Sakaguchi, T., 2007. Refuse patterning and behavioral analysis in a pinniped hunting
camp in the late Jomon Period: a case study in layer V at the Hamanaka 2 site, Re-
bun Island, Hokkaido, Japan. J. Anthropol. Archaeol. 26 (1), 28–46.

UNCORRECTED PROOF
Journal of Anthropological Archaeology xxx (2016) xxx-xxx 11
Scelza, B., 2009. The grandmaternal niche: critical caretaking among Martu Aborig-
ines. Am. J. Hum. Biol. 21, 448–454.
Scelza, B., 2010. Father’s presence speeds the social and reproductive careers of sons.
Curr. Anthropol. 5, 295–303.
Scelza, B., Bliege Bird, R., 2008. Group structure and female cooperative networks in
Australias Western Desert. Hum. Nat. 19, 231–248.
Shennan, S., Downey, S.S., Timpson, A., Edinborough, K., Colledge, S., Kerig, T.,
Manning, K., Thomas, M.G., 2013. Regional population collapse followed initial
agriculture booms in mid-holocene Europe. Nat. Commun. 4, 1–8.
Simms, S.R., 1989. The structure of the Bustos wickiup site, eastern Nevada. J. Calif.
Great Basin Anthropol. 11 (1), 2–34.
Simms, S.R., Heath, K.M., 1990. Site structure of the Orbit Inn: an application of eth-
noarchaeology. Am. Antiq. 55, 797–813.
Smith, E.A., 1991. Inujjuamiut Foraging Strategies: Evolutionary Ecology of an Arctic
Hunting Economy. Aldine de Gruyter, New York.
Speth, J.D., 2010. The Paleoanthropology and Archaeology of Big-Game Hunting.
Springer.
Speth, J.D., Meignen, L.,Bar-Yosef, O., Goldberg, P., 2012. Spatial organization of
Middle Paleolithic occupation X in Kebara Cave (Israel): concentrations of animal
bones. Quatern. Int. 247, 85–102.
Stanner, W.E.H., 1965. Aboriginal territorial organization: estate, range, domain and
regime. Oceania 36, 1–26.
Stevens, N.E., McElreath, R., 2015. When are two tools better than one? Mortars,
millingslabs, and the California acorn economy. J. Anthropol. Ar-
chaeol. 37, 100–111.
Surovell, T.A., 2012. Toward a Behavioral Ecology of Lithic Technology: Cases from
Paleoindian Archaeology. University of Arizona Press.
Tinbergen, N., 1963. On aims and methods in ethology. Z. Tierpsychol. 20, 410–433.
Tipps, B.L., 1993. Investigating the spatial structure of lithic scatter sites from an eth-
noarchaeological perspective: examples from Utah and Nevada. Utah Ar-
chaeol. 6, 57–71.
Tonkinson, R., 1993. The Mardu Aborigines: Living the Dream in Australias Desert,
second ed. Holt, Rinehart & Winston, New York.
Veth, P., Walsh, F., 1988. The concept of “staple”plant foods in the Western Desert
region of Western Australia. Aust. Aboriginal Stud. 2, 19–25.
Walsh, F., 1987. Patterns of Plant Use by Martujarra Aborigines. Master’s thesis, Uni-
versity of Western Australia.
Walsh, F., 2008. To Hunt and to Hold: Martu Aboriginal People’s Uses and Knowl-
edge of Their Country, with Implications for Co-management in Karlamilyi (Rudal
River) National Park and the Great Sandy Desert, Western Australia. PhD thesis,
School of Social and Cultural Studies (Anthropology) and School of Plant Biology
(Ecology), The University of Western Australia.
Whelan, C.S., Whitaker, A.R., Rosenthal, J.S., Wohlgemuth, E., 2013. Hunter-gatherer
storage, settlement and the opportunity costs of women’s foraging. Am. An-
tiq. 78, 662–678.
Williams, A.N., Veth, P., Steffen, W., Ulm, S., Turney, C.S., Reeves, J.M., Phipps,
S.J., Smith, M., 2015. A continental narrative: Human settlement patterns and Aus-
tralian climate change over the last 35,000 years. Quatern. Sci. Rev. 123, 91–112.
Winterhalder, B., Kennett, D.J., Grote, M.N., Bartruff, J., 2010. Ideal free settlement of
California’s Northern Channel Islands. J. Anthropol. Archaeol. 29, 469–490.
Yellen, J., 1977. Archaeological Approaches to the Present. Academic Press, New
York.
Zeanah, D.W., Codding, B.F., Bird, D.W., Bird, R.B., Veth, P.M., 2015. Diesel and
damper: changes in seed use and mobility patterns following contact amongst the
Martu of Western Australia. J. Anthropol. Archaeol. 39, 51–62.