Subsistence Transitions and the Simplification of Ecological Networks in the Western Desert of Australia

Abstract

The Australian desert ecosystem coevolved with humans over the course of fifty millennia, yet our understanding of the place of humans within the ecosystem is only now beginning to deepen; recent research suggests that the removal of Aboriginal people from homelands precipitated rapid ecosystem remodeling. We suggest that network-based approaches are instrumental in broadening our understanding of humans in ecosystems, so apply these approaches to examine nomadic-era ecosystems (when Aboriginal people lived exclusive foraging-based lifeways) and contemporary-era ecosystems (when Aboriginal people live a mixed-based economy lifestyle). Using the approach of food web modeling we explicitly place Martu Aboriginal foragers within the overall ecosystem of the Western Desert. By linking humans to the other biota in the desert, examining each species as nodes in a network and each consumption link as edges in the network, we can better understand the ways the network connectedness shifts between nomadic-era and contemporary-era food webs. Using network randomization simulations we show that the contemporary food webs deviate significantly from the nomadic era food webs, suggesting a key role of humans as “knitters” of the ecosystem. This work has implications for research on resilient ecosystems, both within Australia and beyond, and suggests that humans have significant roles to play in sustainability and resilience.

Full text

1 23
Human Ecology
An Interdisciplinary Journal
ISSN 0300-7839
Hum Ecol
DOI 10.1007/s10745-019-0053-z
Subsistence Transitions and the
Simplification of Ecological Networks in the
Western Desert of Australia
Stefani A.Crabtree, Douglas W.Bird &
Rebecca Bliege Bird

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Science+Business Media, LLC, part of
Springer Nature. This e-offprint is for personal
use only and shall not be self-archived in
electronic repositories. If you wish to self-
archive your article, please use the accepted
manuscript version for posting on your own
website. You may further deposit the accepted
manuscript version in any repository,
provided it is only made publicly available 12
months after official publication or later and
provided acknowledgement is given to the
original source of publication and a link is
inserted to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

Subsistence Transitions and the Simplification of Ecological Networks
in the Western Desert of Australia
Stefani A. Crabtree
1,2,3
&Douglas W. Bird
1
&Rebecca Bliege Bird
1
#Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract
The Australian desert ecosystem coevolved with humans over the course of fifty millennia, yet our understanding of the place of
humans within the ecosystem is only now beginning to deepen; recent research suggests that the removal of Aboriginal people
from homelands precipitated rapid ecosystem remodeling. We suggest that network-based approaches are instrumental in
broadening our understanding of humans in ecosystems, so apply these approaches to examine nomadic-era ecosystems (when
Aboriginal people lived exclusive foraging-based lifeways) and contemporary-era ecosystems (when Aboriginal people live a
mixed-based economy lifestyle). Using the approach of food web modeling we explicitly place Martu Aboriginal foragers within
the overall ecosystem of the Western Desert. By linking humans to the other biota in the desert, examining each species as nodes
in a network and each consumption link as edges in the network, we can better understand the ways the network connectedness
shifts between nomadic-era and contemporary-era food webs. Using network randomization simulations we show that the
contemporary food webs deviate significantly from the nomadic era food webs, suggesting a key role of humans as Bknitters^
of the ecosystem. This work has implications for research on resilient ecosystems, both within Australia and beyond, and
suggests that humans have significant roles to play in sustainability and resilience.
Keywords Food webs .Human-behavioral ecology .Coupled human/natural systems .Australia .Networks
Post-industrial societies are built through large-scale habitat
modification and only thrive due to remotely coupled links
between production and resource consumption (Liu et al.
2016). Such societies support massive per-capita consumption
rates leading to environmental impacts up to 50 times that of
Bplace-based^societies –societies where people live within
the landscapes and ecosystems that support their resource ex-
traction and consumption (Norton 2000; Bliege Bird et al.
2018). Not all post-industrial consumption has negative
environmental impacts, though it is largely responsible for
the rapid ecological changes we face today such as land clear-
ance, pollution, species extinctions, and climate change.
While such changes are characteristic of nearly all continents,
Australia currently has the highest rate of endemic mammal
extinction in the world (Geyle et al. 2018). The principle
agents of Australia’s extinction crisis are invasive mammals
and changing fire-regimes (Burbidge and McKenzie 1989;
Burrows et al. 2006; Woinarski et al. 2011,2015; Ziembicki
et al. 2013,2015;McGregoret al. 2014;Leahyet al. 2016)
consequences of processes linked to settler-colonialism and
changing patterns of consumption.
In Australia’s remote Western Desert these processes are
ongoing. Through the mid-twentieth century massive changes
in fire regimes resulted when Aboriginal people were cleared
from their homelands and patch-mosaic burning for hunting
purposes ceased. Without Indigenous burning, and in the span
of less than three decades, the frequency and seasonality of
fire ignition shifted dramatically, leading to a mean increase in
wildfire size by more than two orders of magnitude (Burrows
et al. 2006). Recent work (Bliege Bird et al. 2012b,2018)has
demonstrated that Aboriginal hunting and burning practices
create diverse niches for native taxa, facilitating the ecological
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s10745-019-0053-z) contains supplementary
material, which is available to authorized users.
*Stefani A. Crabtree
sac376@psu.edu
1
Department of Anthropology, The Pennsylvania State University,
410 Carpenter Building, University Park, PA 16803, USA
2
Center for Research and Interdisciplinarity, 8bis Rue Charles V,
75004 Paris, France
3
Crow Canyon Archaeological Center, 23390 C R K,
Cortez, CO 81321, USA
Human Ecology
https://doi.org/10.1007/s10745-019-0053-z
Author's personal copy

role of species that are themselves functionally vital to arid
ecosystems. This work adds to a growing body of research
suggesting that the loss of such place-based Indigenous live-
lihood practices can cascade through ecosystems by
transforming and simplifying ecological networks, thus con-
tributing to the decline and extinction of vulnerable species
(e.g., Dunne et al. 2016).
To explore the effects of changes in Indigenous livelihoods
on the structure of ecological networks, we ask how the network
structure of human-entangled food webs changes in the contexts
of settler-colonialism and contemporary hybrid economies in
remote Aboriginal Australia. In parts of the Western Desert, a
continued reliance on burning, hunting, and gathering is inte-
grated in complex ways with both private and state sectors of a
globalized economy (Codding et al. 2016). Here we focus on
the trophic interactions in the Martu homelands of the Western
Desert. Martu are the Traditional Owners of 136 k km
2
of
Native Title lands spanning the Great and Little Sandy Deserts
bioregions of Western Australia. Martu were mostly cleared
from their desert homelands by the 1960s when they walked
in or were taken to missions and pastoral stations at the desert’s
fringe. In the 1980s many Martu families returned to their home-
lands, initially to a mostly full-time, but later modified, foraging
economy (Walsh 2008; Zeanah et al. 2015). Using a combina-
tion of interviews and observations of contemporary foraging
strategies, we reconstruct Indigenous livelihoods prior to the
settler-colonial period, examine how the generality of subsis-
tence shifted during the mission era and subsequent re-occupa-
tion, ask whether the structural properties of pre-settler colonial
food webs differ from those of the contemporary period in
exhibiting increased vulnerability to extinction or invasion,
and experimentally consider the effects on network structure
of removing all contemporary human impacts.
Background
The study region is situated within the Little Sandy Desert
bioregion of Australia’s Western Desert. The climate is
semi-arid, with a mean annual rainfall of 363 mm, ranging
from 108 to 1455 mm (Australia 2018a). Temperatures during
the cool/dry winter months (May-Aug) average 10–12 °Cat
night and 25–29 °C during the day; during the hot season
(Sep-Apr), from 34 to 40 °C and 20–26 °Catnight.
Vegetation in the region is mostly (i) spinifex (Triodia s chinzii
and T. basedowii) and acacia (A. pachycarpa and A. ligulata,
among others) on sandplain and dune substrates covering
85.6% of the total land area. Other plant communities include
(ii) lateric uplands and clay-dominated soils with Mulga
(A. aneura) woodland (2.4%) and Senna-dominated shrub-
land (1.1%); (iii) Triodia-dominated but poorly vegetated
rocky ranges (7.3%); and (iv) Eucalyptus (mainly E. victrix
and E. camadulensis)-dominated watercourse margins and
floodplains (3.2%).
The region was inhabited by Martu from Manyjiljarra,
Warnman, and Kartujarra dialect-named groups until 1966–
67, when –in the process of a joint Australian-British program
to establish an intercontinental ballistic missile testing range –
the last groups left for missions and cattle stations on the desert
fringe. Resettlement of Martu homelands began in 1982 with
the establishment of Punmu Community. Parnngurr
Community followed in 1984. Martu have actively hunted
and gathered in this region since then, with many hunting
activities involving the use of patch mosaic burning (Bliege
Bird et al. 2013,2018). In landscapes where Martu hunt more
actively, hunting fires rescale pyrodiversity, preserve access to
long unburned refuge habitat, and buffer against climate-
driven shifts in mean fire size (Bliege Bird et al. 2008,
2012a,2016). Landscapes that are not utilized by Martu ex-
hibit a fire regime with much larger fires and a pronounced
seasonal dominance in ignition toward summer fires. Under an
indigenous fire regime, pyrodiversity boosts populations of
native species such as dingo, monitor lizard, and kangaroo,
even after accounting for mortality due to hunting (Bliege
Bird et al. 2013,2018; Codding et al. 2014).
Prior to European contact, as today, hunting focused on
using fire to expose tracks and dens of smaller animals in turn
creating habitat heterogeneity to support many of the most
important staple plant resources (Jones 1969; Gould 1971;
Kimber 1983; Latz 2004). Martu burned primarily during
the winter months to clear spinifex grassland in order to in-
crease the efficiency of hunting burrowed prey, especially
sand monitor (Varanus gouldii) and other herpetofauna, but
also small mammals such as bilby (Macrotis lagotis, now rare
in the arid zone), mulgara (Dasycercus cristicauda,alsorare),
burrowing bettong (Bettongia lesueur, extinct on the
Australian mainland), and rufous hare wallaby
(Lagorchestes hirsutus, highly endangered). Spot fires for
flushing prey were used while hunting for larger monitors
(V. giganteus and V. panoptes), brushtail possum
(Trichosurus vulpecula, endangered in the Western Desert),
and feral cats. Plant foods were highly seasonal, but grass
seeds such as Eragrostis, Panicum, and Yakirra, and the seeds
of Acacia shrubs were seasonal staples. Many species of
grasses, as well as several bush tomatoes (Solanum centrale
and S. diversiflorum), were only found in great quantities in
recently burnt patches. Finally, while large geophytes are
common in other regions of the desert, Martu country supports
only smaller geophytes such as the pencil yams of Vigna
lanceolata and the corms of Cyperus bulbosus.
In the mid-twentieth century as the Indigenous population
declined and began a process of migration from the desert to
settlements and missions, rapid environmental changes also
occurred. As the influence of people on their ecosystem
waned, invasive species spread, while many species of native
Hum Ecol
Author's personal copy

mammals went into decline or suffered extinction, and bush-
fires increased in extent and intensity (Burrows et al. 2006).
Following re-occupation of Martu homelands in the mid-
1980s, the fire regime was locally restored in a few places
where foragers still maintain frequent access. However, much
of the region is threatened by the spread of both invasive
predators (cats, foxes) and plants such as buffelgrass
(Cenchrus ciliaris). The array of foraged resources has
contracted in the contemporary era to focus mainly on a few
high-ranked species that formed the staple of the pre-contact
era diet (Scelza et al. 2014;Zeanahet al. 2017). Many plant
foods are rarely consumed, but hunting continues to be an
important source of both protein and calories for many
remote-living Martu people, waxing and waning with season-
al shifts in foraging returns, mobility, and cash (Scelza 2012;
Scelza et al. 2014). Overall, purchased foods (mainly less
expensive sources of carbohydrates such as flour, sugar, tea,
and biscuits) comprise about 75% of the diet in the contem-
porary community.
Materials and Methods
Food webs are networks of predator-prey interactions that can
illuminate ecosystem structure and resilience and track how
energy and nutrients flow throughout an ecosystem (Dunne
et al. 2002b; Williams et al. 2002; Yoon et al. 2004). All food
webs require imposed boundaries, limited to particular taxa,
ecological communities, or habitats. The Western Desert food
web created for this work is constructed with a taxonomic
focus on the species consumed by humans and thus is a
human-centered web. We began compiling our data by con-
structing species lists of foods consumed by Martu in both the
nomadic and contemporary eras. Nomadic era species lists
were compiled from ethnographic sources (Gould 1969;
Cane 1987; Latz 2004) and from our own interviews with
foragers who lived during the nomadic period. Our semi-
structured interviews extend back to year 2000, and have de-
veloped in ongoing discussions over many years of living and
working among Martu communities. Since 2000 we have de-
veloped a running list of plants and animals encountered or
listed in regional field guides. Our interviews have been de-
signed to annotate those lists with answers to (at least) the
following questions from all remote-living Martu present in
Parnngurr and Punmu communities who remember foraging
during the nomadic era: Martu taxanomic designation, what
the taxa ate, what ate the taxa, primary habitat (sand plain/
dune, water course woodland, mulga woodland, rocky range,
or cassia upland), and, if exploited by people, the purpose,
foraging methods, and relative importance of the taxon (i.e.,
was the resource acquired nearly every day, regularly, regular-
ly in season, occasionally, or rarely). In many of the semi-
structured interviews as well as in less formal discussions,
Martu have provided rich details on ecological interactions,
animal behavior, and plant/animal life histories.
Contemporary species lists were compiled from Walsh’sdirect
observation of foraging in the Parnngurr region in the years
soon after Martu returned to their homelands (Walsh 1987,
1990,2008; Walsh and Douglas 2011; Walsh et al. 2013),
and from our own quantitative observational records from
4621 forager-hours, during which we participated in foraging
with Martu in the Parnngurr region from 2000 to 2010 (Bliege
Bird and Bird 2008; Bird et al. 2009,2013).
While food webs link species together on a binary scale (a
species is either eaten or not eaten), capturing feeding rate can
offer more descriptive information. To create Fig. 1which
demonstrates the importance of different taxa in Martu diets
we follow methodology (e.g. McCall 2000;Neutelet al.
2002) to estimate the proportion of days that a species was
consumed, then calculate the frequency of consumption on a
target species, summing these values to 100 so they equate to
dietary percentages across each season.
We then compiled the prey and predators of each of the taxa
that were identified as human prey in the sink web of Fig. 1
and connected them via binary feeding links to create the
human-centered food web. In compiling the diet of each or-
ganism, we employ studies from ecosystems that are either
similar to our study area (e.g., King and Green 1979)orover-
lap with our study area (e.g., Codding 2011). To compile the
food web data we began with those taxa that were identified as
food sources for humans. We then identified their predators
and their prey, expanding outward until we ceased identifying
new taxa. This resulted in a list of taxa within a close-knit
consumption community, following protocols developed by
(Dunne et al. 2016; Crabtree et al. 2017b), among others.
All abundant as well as large-bodied organisms were included
in the Martu-centered food web, which is resolved to approx-
imately 12% of all identified species in the Western Desert. A
complete network in this region, including insects, microor-
ganisms, and all plant species, would include approximately
1500 species (Calladine and Crichton 2015).
We identified published manuscripts compiling either gut
content data or highly resolved dietary data from field observa-
tion for each species in this food web (Supplemental Text 1).
Each of these consumption links was then listed for each iden-
tified taxon. If diet was seasonally dependent, we distinguished
between summer vs. winter diets, since that variability may be
important for understanding robustness or vulnerability of taxa
(as summer is the more stressful season); if this level of resolu-
tion was unavailable, then a taxon’s diet was reported the same
across seasons. We focus our analyses here on the summer-
season food webs. A key necessity in compiling these diets is
maintaining the specialist versus generalist feeding patterns of a
taxon (Dunne et al. 2016). While some organisms may switch
to less desired foods when preferred foods are unavailable,
cataloguing all the foods that a taxon might eat in case of famine
Hum Ecol
Author's personal copy

wouldhaveconflationeffects.Therefore, researcher discretion
was used to maintain the specialist versus generalist feeding
patterns of each taxon, tactics used in other studies (Dunne
et al. 2004,2013,2016; e.g., Crabtree et al. 2017b).
The food web compiled here focuses on the diets of verte-
brates. While invertebrates are no doubt key to the function of
the Australian ecosystem, we could not resolve their data
down to the species level, thus grouping them at the Order
level, similar to Crabtree et al. (2017b). In most terrestrial
studies, dietary resolution is easily available for large-bodied
organisms, while invertebrate data can be unevenly resolved.
Consequently, we model invertebrates as a resource, but not as
a consumer, due to conflation effects of highly general feeding
strategies at the Order level, similar to concerns above in
maintaining a generalist versus specialist feeding pattern.
Reconstructing the nomadic era web required dietary infor-
mation from taxa that are currently extinct in our region. To
reconstruct the diets of such species, we used two complemen-
tary datasets: 1) dietary studies in other arid regions (mostly
islands and conservation reserves) where populations can still
be observed, and 2) interviews with Martu women and men
who remember hunting these animals in the 1960s and have
deep ecological understanding of the animals’behavior, life-
history, distribution, and interactions (detailed above). For many
of these small mammals, there are several in-depth reports,
mainly in grey literature, on diet and life-history (Australia
2018b). We use these reports to construct a Bmost likely^dietary
composition for these taxa. During July of 2017 we additionally
interviewed Martu elders regarding diets of each of the extinct
species, checking our data compiled from published reports
against Martu knowledge. For example, Lagorchestus hirsutus
is known to eat primarily monocots in the winter and dicots in
the summer; using this information, we identified which plants
in the Western Desert are monocots and which are dicots and
cross-checked these lists with lists compiled from interviews
with Martu elders to identify its most likely diet.
Our network differs from other, similar studies in that it uti-
lizes dietary observations, not the likelihood of feeding interac-
tion based on body size and modeled encounter rate (Dunne
et al. 2008). While this may result in a more realistic network,
it may also introduce other biases. In particular, observational
data on human diet is likely to be more highly resolved than that
of other animals, due to the fact that there are more observation
hours on humans than on most other desert species. In addition,
humans have the added advantage of being able to report what
they ate and when with a certain amount of fidelity. We
attempted to minimize such biases by using multiple lines of
evidence on the diet of other species (gut content, follow studies,
etc.), and use multiple reports on feeding habits where available,
enabling us to compile highly resolved feeding data for other
species. On the human side, we included only species that were
observed to be consumed, either in the nomadic or contempo-
rary era, or that people reported consuming regularly, with the
result that many low-ranked fallback foods were excluded from
the nomadic era network, even if they were edible and reported
to be consumed rarely. Our estimates of nomadic era diet are
thus likely to err on the side of caution, recording fewer species
interactions than possible.
In the first analysis we calculate the structural properties of
the full food web (Fig. 2) using Network3D (Yoon et al. 2004;
ab
Fig. 1 The constriction of the Martu foraged diet between the nomadic
period (a) and the contemporary period (b) for the summer-season food
web. Here Martu are represented by the silhouette in the center, and each
of their resources are represented radiating around them. Each taxon is
represented by a unique silhouette, and the relative importance in the diet
for Martu is indicated by the width of the connection, with widths
summing to 100. (a) represents 86 consumed taxa, while (b) represents
31 consumed taxa. In (b) the Btaxon^that makes up the largest percent of
the diet is the market economy, here indicated by a sack of flour and a tin
of ham. The simplification of Martu diets is apparent, though some taxa
retain their relative importance between the nomadic period and
contemporary periods, namely sand monitor lizards and solanum fruit
Hum Ecol
Author's personal copy

Williams 2010). A key property, mean path length (MPL), is a
standard measure of network connectivity in food web studies
indicating how many steps on average are required to connect
every taxon in the web. Generally, a network with an MPL of
approximately 2 is a densely connected network; the higher
the score, the less densely connected the network is (Williams
et al. 2002). MPL is formalized as:
Σdij

N
Where dis the distance of node ito node j, here calculated as
the number of edges (feeding links) it takes to connect species
iand every other node (j) in the web. These are then summed
and divided by the number of species represented in the food
web. Mean path length has been empirically shown to be a
reliable measure for examining how quickly effects, such as
increases in population of a species or extinction events,
would be felt throughout the web (Williams et al. 2002;
Dunne et al. 2016;Crabtreeet al. 2017b).
Additional network metrics described here include those that
indicate how many taxa are producers (basal), how many are
both consumers and consumed by other species (intermediate)
and how many taxa only consume but are not consumed by other
taxa (top). To assess degree distributions of the number of links
per consumer, we use the scaling relationship between taxa and
the number of prey they consume (Fig. 3)(Dunneet al. 2016;
Crabtree et al. 2017a). We also examine the percentage of her-
bivorous species (taxa that only consume basal nodes); each of
these measures scales from 0 to 1. Additionally we examine the
average short weighted trophic level of the web as food webs that
are Btop heavy^(indicating many taxa consuming at the highest
levels) tend toward instability (Dunne et al. 2002b). Finally, we
examine vulnerability, connectivity, and generality standard de-
viations which report variance around the mean of how vulnera-
ble a taxon is to predation, how generalist or specialist of a con-
sumption strategy a taxon has, and the combination of the two
which examines how connected in general taxa are. Connectivity
and path length therefore can be taken together to examine how
quickly effects would propagate throughout a food web and how
central a species is within a highly connected network. Results
calculated from the full food web are reported in Table 2,and
additional metrics are reported in Supplemental Table 1.
To determine the significance of our measures of network
structure, we cannot simply compare our network with pub-
lished networks, as the quantity of links and edges in a food
web will change the results for specific network statistics
(Romanuk et al. 2009,2017; Thompson et al. 2012;Dunne
Fig. 2 The full Martu food web, representing 173 unique taxa with 1149
feeding links. The basal nodes in red represent primary producers (plants)
while each increasing trophic level is indicated by higher nodes in lighter
colors; true yellow nodes are true carnivores. Edges represent feeding
links and they attenuate toward the resource; looped edges indicate
cannibalism. This representation was created in the open source
software Network3D (Yoon et al. 2004; Williams 2010)
Camel, Crow, and Python
Humans
1
2
3
4
5
6
7
8
0255075100
Number of resources
Number of consumer taxa
Histogram of number of resource taxa per consumer
Ramsay’s Python and Bustard
Cat
and Dingo
Fig. 3 Histogram of number of
resource taxa per consumer,
similar to Dunne et al. 2016
(Fig. 3). With 86 consumed taxa
humans are Bsuper generalists.^
Their removal causes a less
skewed and more compressed
distribution of feeding links
Hum Ecol
Author's personal copy

et al. 2013). Significance is commonly addressed through
simulating random networks with the same number of links
and nodes. Our random models employ a Monte Carlo simu-
lation extending an Erdös-Renyi randomization process (Zhu
et al. 2014), creating random networks that maintain the tro-
phic structure of the food web, a method known as the Niche
Model (Dunne et al. 2008,2013; Romanuk et al. 2017). The
randomization process uses two input parameters, species
richness and the number oflinks, and parameterizes the model
with the same number of nodes and links. We then assign
feeding links between two nodes, using maximum likelihood
methods to parameterize the model from the food web data
(Romanuk et al. 2017). These methods identify the likely
structural properties of food webs with similar number of taxa
and similar number of feeding links. Calculating the model
mean and standard deviation from each of the 1000 runs pro-
duces a 95% confidence interval for various food web struc-
tural values, results reported in Supplemental Table 2. We then
compare the modern food web data to these confidence enve-
lopes to determine the extent to which empirical network
structure deviates from expected values.
Finally, to explore the structure of feeding networks with-
out humans present, as would be the case for much of the
Western Desert today where people are essentially absent
and as was the case for the 20+ years that Martu were in
missions and pastoral stations, we simulate the contemporary
food web without Martu feeding links. This dataset includes
all extant modern taxa, removing humans as both predator and
prey, which is accomplished through node knock-out simula-
tions (see Crabtree et al. 2017b).
Results
Nomadic Vs. Contemporary Martu Resource
Selectivity
The complete nomadic era food web network includes 173
unique taxa with 1149 feeding links among them. In the no-
madic era summer network, Martu directly consumed 86 taxa
compared to the modern network’s 31 taxa. During the no-
madic era, well over half of the represented taxa were at least
occasionally consumed by human foragers; today, around
17% are occasionally consumed. However, the contemporary
taxa with the strongest past interactions are those that are also
characterized by strong present interactions: namely monitor
lizards, solanum fruit, grasses, and acacia. Several species
with strong past interactions are no longer consumed because
they are locally extinct, including Lagorchestes hirsutus and
Bettongia leuseur.
The human node-specific metrics show a diminution of the
central role of Martu between the nomadic and contemporary
webs. Martu generality scores decrease substantially, as does
connectivity (Table 1). Martu path length increases, indicating
the decreasing centrality of humans in this web. While their
short weighted trophic level increases slightly in the modern
web, this is likely a reflection of the aggregate nature of the
market economy node. The vulnerability of Martu to preda-
tion decreases, while the clustering coefficient (a measure of
mutual predation of Martu prey) also decreases.
Nomadic Vs. Contemporary Network Structure
The generality standard deviation, which measures variability
in the number of prey, changes substantially between the no-
madic era and the modern era, with much higher generality in
the nomadic era. This reflects the simplification in human
diets, as they exhibit a greater number of prey during the
nomadic era (Fig. 3) than any other taxa. The Top Taxa mea-
surement also shows a marked difference between the nomad-
ic era and today. In the nomadic era there are no taxa that are
not consumed by others, while in the modern data several
venomous snakes do not have regular predators. The slight
decline in herbivorous taxa between the nomadic era and to-
day indicates the local extinction of several small-bodied
mammals; this result is also reflected in the declining percent-
age of intermediate taxa. As the small-bodied mammals were
both important prey for a variety of species in the region and
also were important consumers of many plant species, their
loss can be detected in these structural properties.
The Distribution of Feeding Links
As with the networks generated for the Sanak archipelago
(Dunne et al. 2013,Fig.3) our nomadic era webs show
humans at the very tail end of a right skewed distribution of
feeding links (Fig. 3). The distribution shifts substantially
from the nomadic to the contemporary webs; if Martu are
removed from Fig. 3the most connected taxa include camels,
crows, and pythons with around 35 feeding links each, while
Table 1 Node level statistics for Martu foragers, with comparison between nomadic and modern era food webs. Each of the statistics demonstrates the
movement of Martu from being a central node in the Western Desert to a more peripheral node
SWTL Generality Vulnerability Connecivity Path Clustering
Nomadic 2.348 12.949 0.904 6.926 1.506 0.054
Contemporary 2.372 5.167 0.833 3 1.905 0.047
Hum Ecol
Author's personal copy

in general the contemporary web has a less skewed feeding
link distribution, indicating a more equal distribution of prey
among consumers.
Empirical Vs. Random Network Structure
Finally, to examine whether the changes in metrics between
the nomadic and modern era food webs are statistically sig-
nificant, we utilize the output from 1000 Monte Carlo simu-
lations of food webs with the same input parameters of num-
ber of species and number of links. Figures 4and 5report the
standard deviation around the mean, the empirical values from
the nomadic era and the modern era, as well as the simulated
contemporary dataset that removes humans. Changes in per-
cent of trophic level measures are not statistically significant
and could be predicted based on the nomadic priors (Fig. 4a).
The dramatic decrease in the number of links per species be-
tween the nomadic era and the modern era cannot be predicted
based on priors, and therefore the decrease is statistically sig-
nificant (Fig. 4b). Changes in generality standard deviation,
mean path length, the number of links, and the clustering
coefficient are all statistically significant and could not be
predicted based on priors (Fig. 5a, b). Concomitantly, for all
of the values that the contemporary data performed poorly,
they performed even poorer with the node knock-out simula-
tion dataset, e.g., the decrease in the number of links per spe-
cies, as well as changes in generality standard deviation, mean
path length and clustering coefficient all perform more poorly
without Martu. Other metrics for both the contemporary
dataset and the simulated no-Martu dataset, however, fall
within predicted variability given the priors. These values
are shown in Table 2and in Figs. 4and 5,withimplications
discussed below.
Discussion
Ecological Impacts of Declining Generalism
Our results suggest that changes to network connectivity with
market integration or subsistence intensification may have the
potential to lead to substantial ecological changes. In the
Western Desert, ecological changes are evident in the differ-
ences between the contemporary and nomadic era networks.
Between the time span represented by the nomadic era web
and the contemporary web, 10 species of small mammal went
extinct, 14 mammals are currently threatened, as are 3 bird
species and 2 reptile species (Australia 2009; Geyle et al.
2018). Invasive species became common, especially camels,
cats, foxes, and donkeys. The changes to network structure—
the spread of invasive species and the extinction of others—
are consistent with the loss of a super-generalist that per-
formed an important functional role in the ecosystem
(Dunne et al. 2016). While Martu returned to the system in
the mid 1980s, the importance of their functional role has
declined with the loss of some of their dietary generality,
0.0
0.2
0.4
0.6
% Basal
% Herbivorous
% Intermediate
% Omnivorous
% Top
Scores
Trophic Levels Comparison Data
6.0
6.2
6.4
Links / species
Links per species
Fig. 4 Comparison of simulated Niche model output to empirical data
and node knock-out data. Error bar corresponds to the niche model
output; the rust colored circle corresponds to the nomadic era food web
statistics; the brown triangle corresponds to the contemporary food web
statistics; the yellow diamond corresponds to the simulated food web with
no Martu (node knock-out simulation). In (a) we see that nomadic,
contemporary, and simulated data all fall within Niche model predicted
variability, indicating remarkable structural stability of the food web
despite extinctions. In (b) however we see that the number of links per
species decreases dramatically between the nomadic era (rust circle) and
both the contemporary and simulated data, which fall well outside of the
error bars. This indicates that the loss of Martu as Bsuper generalists^is
statistically significant and would not be predicted based on regular
network fluctuation. Moreover, the relatively poorer performing
simulated data indicate that even with less of an effect than during
nomadic times, Martu still perform important functions within the
ecosystem
Hum Ecol
Author's personal copy

though of all species, they still remain the most central to the
structure and function of the contemporary network.
Theoretical work in ecology suggests that the structure of
interaction networks affects the long-term stability of species
assemblages (May 1972; McNaughton 1978; Pimm 1979;
McCann et al. 1998; Neutel et al. 2002). Stability is often
conceptualized as resilience in the face of extinction—e.g.,
few secondary losses caused by the initial loss of one species
(Solé and Montoya 2001; Terborgh et al. 2001;Dunneet al.
2002a;Memmottet al. 2004; Srinivasan et al. 2007;Rezende
et al. 2007), high levels of integration and persistence in the
face of external perturbations (Solé and Montoya 2001;
Dunne and Williams 2009), and fewer changes to species
compositions and interactions in the face of invasive organ-
isms (Post and Pimm 1983; Bartomeus et al. 2008). Networks
that have a high number of links (high connectance) stabilize
ecosystem processes in variable conditions, while those with a
preponderance of weak connections between species (many
generalists, and few specialists) tend to be more resistant to
invasion (Tylianakis et al. 2010).
While we show that the basic structure of the network does
not change substantially from the nomadic to the contempo-
rary era—indicative of remarkable structural stability (Fig.
4a)—there are nonetheless large shifts in connectance (Figs.
4band5). This shift in connectance is mirrored in the feeding
link distribution (Fig. 3) and has important implications for
network robustness. In network studies, those nodes with the
highest connectivity have been shown to confer high stability
in networks, helping them resist extinctions (Barabási and
Albert 1999) but with their removal, networks dramatically
rearrange. Overall, food webs with high connectance domi-
nated by a few super-generalists are better suited to resist
invasive species, but are more sensitive to the loss of those
highly connected species (Romanuk et al. 2009;Baiseret al.
2010; Smith-Ramesh et al. 2017). In our networks, Martu are
the most generalist node, consistent with other empirical webs
Table 2 Common network statistics for food webs. Path is mean path
length, a standard measure of network connectance. Basal is the percent
of the web representing primary producers. Intermediate are those taxa
that both consume other taxa and are consumed. Top are those taxa that
have no predators. Generality SD is the generality standard deviation, a
measure of how general of a feeding strategy the web exhibits.
Vulnerability standard deviation is a measure of how vulnerable on
average taxa are to predation. Trophic Level is the average trophic level
for the web—a normal trophic level is around 1.5. Percent herbivorous
are the percent of taxa that consume only plants. These are the standard
measures that help describe food web structure and have been used in
many other publications (Dunne et al. 2002b; Williams et al. 2002;
Dunne 2006;Dunneet al. 2016)
Era Mean Path Length Basal Intermed Top Generality SD Vuln SD Conn SD Average Trophic Level % Herbiv
Nomadic 2.459 0.549 0.451 0 1.640 1.041 0.89 1.511 0.266
Modern 2.654 0.575 0.418 0.018 1.503 1.088 0.842 1.498 0.247
No Humans 2.704 0.563 0.402 0.036 1.494 1.119 0.835 1.498 0.251
1.0
1.5
2.0
2.5
Generality SD
Mean Path Length
Vulnerability SD
Scores
Path, Generality, and Vulnerability
0.00
0.25
0.50
0.75
1.00
# of Links SD
Clustering Coeff
Connectance
Max Similarity
Connectivity Statistics
Fig. 5 Comparison of simulated Niche model output to empirical data
and node knock-out data. Error bar corresponds to the niche model
output; the rust colored circle corresponds to the nomadic era food web
statistics; the brown triangle corresponds to the contemporary food web
statistics; the yellow diamond corresponds to the simulated food web with
no Martu (node knock-out simulation). In (a) we see that contemporary
and simulated MPL and Generality are statistically outside of predicted
variability given nomadic-era priors. In (b) we see that the number of
links standard deviation and clustering coefficient are also statistically
deviant from expected variability. The other metrics—vulnerability to
predation, connectance, and maximum similarity—however, fall within
predicted variability further demonstrating the structural consistency of
the overall food web despite extinctions
Hum Ecol
Author's personal copy

suggesting humans as ‘super generalists,’(Dunne et al. 2016;
Crabtree et al. 2017b). When Martu are removed, as we sim-
ulate in our node knock-out analysis, the most connected taxa
include crows, pythons, and camels (see Fig. 3). Our results
suggest that the Western Desert food web became more sus-
ceptible to invasions following the removal of Martu to out-
stations, which may have precipitated extinctions of small-
bodied mammals by facilitating the invasion of feral predators
and other species (e.g., camels).
Our results are consistent with other studies showing that
with increasing sedentism the number of wild taxa incorporat-
ed into the diet typically narrows, and presumably so too do
other forms of nonconsumptive species interactions such as
gathering for clothing production (Maccord and Begossi
2006; Bharucha and Pretty 2010). Similar patterns of ecolog-
ical simplification have been observed in the Philippines
among the Batak, where contemporary foraging camps are
smaller and shorter in duration and foragers take a narrower
range of forest products (Eder 1988). In Brazil, urbanization
and the change from artisanal methods to commercially based
cultivation reduces the diversity of species taken in coastal
fishing villages (Hanazaki and Begossi 2003; Maccord and
Begossi 2006), increasing the dependence of local populations
on externally purchased products and altering their relation-
ship with local resources.
Other empirical food webs demonstrate rapid change fol-
lowing the loss of ‘super generalist’human foragers. For ex-
ample, Aleut harvesting of seals (for kayak skins) influenced
stable cod populations (Maschner et al. 2009;Dunneet al.
2016) while with the removal of Aleut from traditional hunt-
ing grounds cod populations decreased in part due to seals
losing a primary predator. In coastal Chile, human predation
on carnivorous gastropods and grazing limpets increased the
density of mussels and the recruitment of limpets (Castilla
1999) while the later exclusion of human subsistence
shellfishing from intertidal zones in Chile caused an over-
abundance of predatory gastropods and grazing limpets, dev-
astating mussel beds, increasing barnacles, and reducing
macroalgal cover (Castilla 1999). In these cases, the removal
of humans from food webs caused networks to adapt as the
strength and diversity of human interactions lessened.
Processes Influencing Ecological Simplification
Changes to generality implicated in the shift from foraging to
a market economy may stem from a variety of often
interacting processes. First, reductions in residential mobility
that are a hallmark of colonial interactions play an important
role in the degree of dietary diversity among nomadic popu-
lations. Models of mobility derived from foraging theory
(Orians and Pearson 1979;Schoener1979) predict that for-
agers will increasingly specialize on higher quality food
patches as distance from the central place increases, because
the total rate of energy acquisition must incorporate the travel
and transport costs associated with foraging in a more distant
patch. This shift toward central place foraging has substantial
effects on the ecosystem—as generality decreases in the food
web modeled here, connectivity also decreases, leading to a
more fragmented food web that is susceptible to invasions.
This reduction in mobility might also result in a shift from
broad impacts on the environment to more intensive local
exploitation and environmental homogenization due to a re-
duced scale of vegetation disturbance or increased fire sup-
pression (Bliege Bird and Nimmo 2018). Increased environ-
mental homogeneity also tends to favor more specialized for-
aging tactics (Futuyma and Moreno 2014)aswellasreducing
the total number of species available to acquire. This may also
lead to dietary contraction, which is often thought to be a
function of the reduced risk of hunger provided by access to
purchased foods (Eder 1988).
Transport availability, such as the introduction of motor-
ized vehicles, also affects the array of resources selected.
When Cree began to use snowmobile and outboard motors,
diet generality contracted to fewer, highly ranked species
(Winterhalder 1981). As Zeanah et al. (2015) show, contrac-
tions in selectivity can occur in two situations. First, if foragers
must travel long distances between patches, resources that
require long processing times may drop out due to tradeoffs
with time needed to travel. Second, as Winterhalder (1981)
suggested, since motor vehicles increase the distance that can
be traveled in a given amount of time, encounter rates (number
per unit time) with higher ranked resources also increase, and
foragers face tradeoffs between traveling to a new foraging
location or staying and accepting lower foraging returns.
However, due to depletion around sedentary, permanent com-
munities, foragers may need to travel farther from the central
place over time (Hames and Vickers 2018).
In the Martu case in particular, the most salient change in
selectivity has come about as a result of the dropping of wild
grass and tree seeds from the diet. Prior to European contact,
these were the most constant staple resource commonly har-
vested, and formed the basis of much of the pre-contact econ-
omy throughout arid Australia (O’Connell and Hawkes 1984).
As Zeanah et al. (2015) demonstrate, for Martu this change in
selectivity is not simply a substitution of foraged resources for
commercial food. The availability of 4-WD vehicles and die-
sel fuel have altered the cost-benefits of contemporary Martu
foraging, making seed harvesting uneconomical by imposing
a conflict between time spent processing seeds and travel time.
This opportunity cost is incurred because of the processing
costs of seeds after harvesting; today, a woman foraging in
circumstances where harvesting seeds would improve her
overall foraging return rate faces a choice between spending
the rest of the day processing seeds at a foraging camp, or
traveling by vehicle either back to a community (where there
are opportunities for alternative social and economic
Hum Ecol
Author's personal copy

activities), or traveling to a new resource patch (for example to
hunt for sand monitor or bustard). As a result, resources such
as sand monitor and bustard remain important in Martu food
webs while seeds are not.
With economic changes also come greater opportunity
costs of foraging due to the need to invest in alternative eco-
nomic activities. This may manifest in a desire to minimize
energetic expenditure, as with the Semaq Beri of Malaysia
who have preferred purchased foods because they can be ac-
quired with less energetic effort than foraged foods
(Kuchikura 1988). Spatial contraction of foraging to easily
accessible regions due to these opportunity costs may further
limit dietary diversity if there is spatial heterogeneity in spe-
cies distributions.
The introduction of new technologies often changes the
costs and benefits of acquiring a wider array of resources.
Some technologies (e.g., rifles) may democratize foraging
leading to fewer differences among individuals and a reduced
degree of generality variation across individuals (Svanbäck
and Bolnick 2007).Butaccesstotechnologytosupportfor-
aging often limits participation in contemporary ‘mixed econ-
omies’leading to a positive relationship overall between ac-
cess to wealth and dietary diversity (Godoy et al. 2005; Scelza
et al. 2014;ReadyandPower2017). Furthermore, as Scelza
et al. (2014) point out, dietary diversity may be maintained if
purchased foods are not substitutable with foraged foods due
to the alternative cultural values associated with acquiring or
sharing such foods.
Conclusions
We propose that humans played a key ecological role in the
function of Western Desert food webs, and we demonstrate
that dietary changes following sedentism and changes in mo-
bility patterns decreased the number of species interactions
and potentially contributed to the comparative simplification
of the contemporary food web. We show that while the basic
structure of the Western Desert food web remains the same
between the nomadic and contemporary eras, there are sub-
stantial and significant changes in connectivity, indicating the
key role humans played (and continue to play) in the function
of this ecosystem. This has important implications for ecosys-
tem health and suggests that a nuanced (networked) under-
standing of the human place in species interactions is key to
understanding resilience.
A growing body of research suggests that humans per-
form key ecological functions in ecosystems (Dunne et al.
2016; e.g., Bliege Bird and Nimmo 2018); our work sup-
ports these findings, suggesting that the roles that humans
perform in many ecosystems may lead to more stable in-
teractions that are resistant to invasions by non-native or-
ganisms. This is importantly a function of the extreme
generality of many human foraging strategies. Thus, shifts
in selectivity resulting from the processes of globalization,
sedentism, and technological, social and economic change
have profound ecological effects that may result in simpli-
fied, less connected ecological networks. The contraction
of hunting effort to a few species reduces rates of predation
on others, which can lead to mesopredator release
(Sutherland et al. 2011)—a surge in the population and
predation of smaller predators which in turn puts their prey
at risk of extinction (Crooks and Soulé 1999;Ziembicki
et al. 2015;Leahyet al. 2016). Shifts in diet from wild to
purchased foods remove people from their local networks
almost entirely, and since, as we show, people were for-
merly highly connected generalists, non-native species
may have more easily invaded (Romanuk et al. 2009,
2017). But consumption effects are not the only important
influence that people have on ecological networks: equally
important may be the nonconsumptive effects of ecosystem
engineering. If shifts in selectivity involve dropping re-
sources that formerly were associated with substantial eco-
system engineering effects, such as resources that involve
small-scale vegetation clearing, digging or other bioturba-
tion activities, hydrological engineering (such as the pro-
visioning of water sources or wetlands), or the use of land-
scape fire (e.g. burning to improve hunting returns), we
might also suspect substantial ecological changes.
Previous work in our region shows that the loss of small-
scale human-lit hunting fires causes wildland fire size to
dramatically increase, reducing the structural diversity of
vegetation at the landscape scale (Burrows et al. 2006;
Bliege Bird et al. 2012a,b). This in turn may have had a
profound effect on the survival of many native species, and
increased the predation efficiency of invasive predators
(e.g., cats, foxes) (Bliege Bird et al. 2018).
This work highlights the importance of modeling humans
within ecosystems, as most ecosystems worldwide have expe-
rienced a deep coevolutionary history with humans. By
modeling the effects that people have on ecosystems, we will
be able to better understand the multifaceted ways that
humans impact environments, both positively and negatively.
Acknowledgments First and foremost our gratitude goes to all of our
Martu colleagues, friends, and family that have made this work possible.
This work has been generously supported by grants from the National
Science Foundation (BCS-1459880) and the Max Planck Institute for
Evolutionary Anthropology.
Compliance with Ethical Standards
Conflict of Interest The authors declare that they have no conflict of
interest.
Publisher’sNoteSpringer Nature remains neutral with regard to jurisdic-
tional claims in published maps and institutional affiliations.
Hum Ecol
Author's personal copy

References
Australia, Commonwealth of (2009) Assessment of Australia’s terrestrial
biodiversity 2008. Current
Australia, Commonwealth of (2018a) Australian Government : Bureau of
Meteorology. http://www.bom.gov.au. Accessed 5 Jan 2018
Australia, Commonwealth of (2018b) Mala/rufous hare-wallaby. In:
Aust. Wildl. Conserv. http://www.australianwildlife.org/wildlife/
mala-rufous-hare-wallaby.aspx. Accessed 5 Jan 2018
Baiser, B., Russell, G. J., and Lockwood, J. L. (2010). Connectance
determines invasion success via trophic interactions in model food
webs. Oikos 119: 1970–1976. https://doi.org/10.1111/j.1600-0706.
2010.18557.x.
Barabási, A.-L., and Albert, R. (1999). Emergence of scaling in random
networks. Science 286(80): 509 LP–509512.
Bartomeus, I., Vilà, M., and Santamaría, L. (2008). Contrasting effects of
invasive plants in plant–pollinator networks. Oecologia 155: 761–
770. https://doi.org/10.1007/s00442-007-0946-1.
Bharucha, Z., and Pretty, J. (2010). The roles and values of wild foods in
agricultural systems. Philos trans R Soc B biol Sci 365: 2913 LP–
2912926.
Bird, D. W., Bird, R. B., and Codding, B. F. (2009). In pursuit of Mobile
prey: Martu hunting strategies and Archaeofaunal interpretation.
Am Antiq 74: 3–29. https://doi.org/10.1017/S000273160004748X.
Bird, D. W., Codding, B. F., Bliege Bird, R., et al (2013). Megafauna in a
continent of small game: Archaeological implications of Martu cam-
el hunting in Australia’s Western Desert. Quat Int 297: 155–166.
https://doi.org/10.1016/j.quaint.2013.01.011.
Bliege Bird, R., and Bird, D. W. (2008). Why women hunt: Risk and
contemporary foraging in a Western Desert aboriginal community.
Curr Anthropol 49: 655–693. https://doi.org/10.1086/587700.
Bliege Bird, R., and Nimmo, D. (2018). Restore the lost ecological func-
tions of people. Nat Ecol Evol. https://doi.org/10.1038/s41559-018-
0576-5.
Bliege Bird, R., Bird, D. W., Codding, B. F., et al (2008). The Bfire stick
farming^hypothesis: Australian aboriginal foraging strategies, bio-
diversity, and anthropogenic fire mosaics. Proc Natl Acad Sci 105:
14796–14801.
Bliege Bird, R., Codding, B., Kauhanen, P., and Bird, D. (2012a).
Aboriginal hunting buffers climate-driven fire-size variability in
Australia’s spinifex grasslands. Proc Natl Acad Sci 109: 10287–
10292. https://doi.org/10.1073/pnas.1204585109.
Bliege Bird, R., Codding, B. F., Kauhanen, P. G., and Bird, D. W.
(2012b). Aboriginal hunting buffers climate-driven fire-size vari-
ability in Australia’s spinifex grasslands. In Proc Natl Acad Sci
109:10287 LP-10292.
Bliege Bird, R., Tayor, N., Codding, B. F., and Bird, D. W. (2013). Niche
construction and dreaming logic: Aboriginal patch mosaic burning
and varanid lizards (Varanus gouldii) in Australia. Proc R Soc B Biol
Sci 280: 20132297. https://doi.org/10.1098/rspb.2013.2297.
Bliege Bird, R., Bird, D. W., and Codding, B. F. (2016). People, El Niño
southern oscillation and fire in Australia: Fire regimes and climate
controls in hummock grasslands. Phil Trans R Soc B 371:
20150343.
Bliege Bird, R., Bird, D. W., Fernandez, L. E., et al (2018). Aboriginal
burning promotes fine-scale pyrodiversity and native predators in
Australia’s Western Desert. Biol Conserv 219: 110–118. https://
doi.org/10.1016/j.biocon.2018.01.008.
Burbidge, A. A., and McKenzie, N. L. (1989). Patterns in the modern
decline of Western Australia’s vertebrate fauna: Causes and conser-
vation implications. Biol Conserv 50: 143–198.
Burrows, N. D., Burbidge, A. A., Fuller, P. J., and Behn, G. (2006).
Evidence of altered fire regimes in the Western Desert region of
Australia. Conserv Sci West Aust 5: 14–26.
Calladine, A., and Crichton, E. (2015). Atlas of living Australia science
symposium, Review of Online and Desktop Tools for the ALA, In,
pp. 1–309.
Cane, S. (1987). Australian aboriginal subsistence in the Western desert.
Hum Ecol 15: 391–434. https://doi.org/10.1007/BF00887998.
Castilla, J. C. (1999). Coastal marine communities: Trends and perspec-
tives from human-exclusion experiments. Trends Ecol Evol 14:
280–283. https://doi.org/10.1016/S0169-5347(99)01602-X.
Codding BF (2011) BAny kangaroo?^On the ecology, ethnography and
archaeology of foraging in Australia’s arid west. Stanford University
Codding, B. F., Bird, R. B., Kauhanen, P. G., and Bird, D. W. (2014).
Conservation or co-evolution? Intermediate levels of aboriginal
burning and hunting have positive effects on kangaroo populations
in Western Australia. Hum Ecol 42: 659–669. https://doi.org/10.
1007/s10745-014-9682-4.
Codding, B. F., Bliege Bird, R., Bird, D. W., and Zeanah, D. W. (2016).
Alternative aboriginal economies: Martu livelihoods in the 21st cen-
tury. In Codding, B. F., and Kramer, K. (eds.), Why forage? 21st
century hunting and gathering. University of new Mexico Press, pp.
185–211.
Crabtree, S. A., Bocinsky, R. K., Hooper, P. L., et al (2017a). How to
make a polity (in the Central Mesa Verde region). Am Antiq 82: 71–
95. https://doi.org/10.1017/aaq.2016.18.
Crabtree, S. A., Vaughn, L. J. S., and Crabtree, N. T. (2017b).
Reconstructing Ancestral Pueblo food webs in the southwestern
United States. J Archaeol Sci 81: 116–127. https://doi.org/10.
1016/j.jas.2017.03.005.
Crooks, K. R., and Soulé, M. E. (1999). Mesopredator release and avi-
faunal extinctions in a fragmented system. Nature 400: 563.
Dunne, J. A. (2006). The network structure of food webs. In Pascual, M.,
and Dunne, J. A. (eds.), Ecological networks: Linking structure to
dynamics in food webs, Oxford University Press, New York, USA,
pp. 27–85.
Dunne, J. A., and Williams, R. J. (2009). Cascading extinctions and
community collapse in model food webs. Philos Trans Biol Sci
364: 1711–1723.
Dunne, J. A., Williams, R. J., and Martinez, N. D. (2002a). Network
structure and biodiversity loss in food webs: Robustness increases
with connectance. Ecol Lett 5: 558–567. https://doi.org/10.1046/j.
1461-0248.2002.00354.x.
Dunne, J. A., Williams, R. J., and Martinez, N. D. (2002b). Network
structure and biodiversity loss in food webs: Robustness increases
with connectance. Ecol Lett 5: 558–567. https://doi.org/10.1046/j.
1461-0248.2002.00354.x.
Dunne, J. A., Williams, R. J., and Martinez, N. D. (2004). Network
structure and robustness of marine food webs. Mar Ecol Prog Ser
273: 291–302. https://doi.org/10.3354/meps273291.
Dunne, J. A., Williams, R. J., Martinez, N. D., et al (2008). Compilation
and network analyses of Cambrian food webs. PLoS Biol 6: 693–
708. https://doi.org/10.1371/journal.pbio.0060102.
Dunne, J. A., Lafferty, K. D., Dobson, A. P., et al (2013). Parasites affect
food web structure primarily through increased diversity and com-
plexity. PLoS Biol 11. https://doi.org/10.1371/journal.pbio.
1001579.
Dunne, J. A., Maschner, H., Betts, M. W., et al (2016). The roles and
impacts of human hunter-gatherers in North Pacific marine food
webs. Sci Rep 6: 21179.
Eder, J. F. (1988). Batak foraging camps today: A window to the history
of a hunting-gathering economy. Hum Ecol 16: 35–55. https://doi.
org/10.1007/BF01262025.
Futuyma, D. J., and Moreno, G. (2014). The evolution of ecological
specialization. Annu Rev Ecol Syst 19: 207–233.
Geyle, H. M., Woinarski, J. C. Z., Baker, G. B., et al (2018). Quantifying
extinction risk and forecasting the number of impending Australian
bird and mammal extinctions. Pacific Conserv Biol 24: 157–167.
https://doi.org/10.1071/PC18006.
Hum Ecol
Author's personal copy

Godoy, R., Reyes-García, V., Byron, E., et al (2005). The effect of market
economies on the well-being of indigenous peoples and on their use
of renewable natural resources. Annu Rev Anthropol 34: 121–138.
https://doi.org/10.1146/annurev.anthro.34.081804.120412.
Gould, R. A. (1969). Subsistence behaviour among the Western Desert
aborigines of Australia. Oceania 39: 253–274. https://doi.org/10.
1002/j.1834-4461.1969.tb01026.x.
Gould, R. A. (1971). Uses and effects of fire among the Western Desert
aborigines of Australia. Mankind 8: 14–24.
Hames, R. B., and Vickers, W. T. (2018). Optimal dietbreadth theory as a
model to explain variability in Amazonian hunting. Am Ethnol 9:
358–378. https://doi.org/10.1525/ae.1982.9.2.02a00090.
Hanazaki, N., and Begossi, A. (2003). Does fish still matter? Changes in
the diet of two Brazilian fishing communities. Ecol Food Nutr 42:
279–301. https://doi.org/10.1080/03670240390229643.
Jones, R. (1969). Fire-stick farming. Aust Nat Hist 16: 224–228.
Kimber, R. (1983). Black lightning: Aborigines and fire in Central
Australia and the Western Desert. Archaeol Ocean 18: 38–45.
King, D., and Green, B. (1979). Notes on diet and reproduction of the
sand goanna , Varanus gouldii rosenbergi. Copeia 1979: 64–70.
Kuchikura, Y. (1988). Food use and nutrition in hunting and gathering
Community in Transition, peninsular Malaysia. Man Cult Ocean 4:
1–30.
Latz P (2004) Bushfires and Bushtucker: Australian aboriginal plant use
in Central Australia. IAD Press
Leahy, L., Legge, S. M., Tuft, K., et al (2016). Amplified predation after
fire suppresses rodent populations in Australia’s tropical savannas.
Wild l Res 42: 705–716. https://doi.org/10.1071/WR15011.
Liu, J., Yang, W., and Li, S. (2016). Framing ecosystem services in the
telecoupled Anthropocene. Front Ecol Environ 14: 27–36. https://
doi.org/10.1002/16-0188.1.
Maccord PL, Begossi A (2006) Dietary Changes over Time in a Caiçara
Community from the Brazilian Atlantic Forest. Mudanças dietéticas
com o passar do tempo em uma Comunidade de Caiçara da Floresta
Atlântica brasileira. Ecol Soc 11:13
Maschner, H. D. G., Betts, M. W., Cornell, J., et al (2009). An introduc-
tion to the biocomplexity of Sanak Island , Western Gulf of Alaska.
Pacific Sci 63: 673–709.
May, R. M. (1972). Will a large complex system be stable? Nature 238:
413.
McCall, G. S. (2000). Ju/‘hoansi adaptations to a cash economy. African
Sociol Rev 4: 138–155.
McCann, K., Hastings, A., and Huxel, G. R. (1998). Weak trophic inter-
actions and the balance of nature. Nature 395: 794–798. https://doi.
org/10.1038/27427.
McGregor, H. W., Legge, S., Jones, M. E., and Johnson, C. N. (2014).
Landscape management of fire and grazing regimes alters the fine-
scale habitat utilisation by feral cats. PLoS One 9. https://doi.org/10.
1371/journal.pone.0109097.
McNaughton, S. J. (1978). Stability and diversity of ecological commu-
nities. Nature 274: 251.
Memmott, J., Waser, N. M., and Price, M. V. (2004). Tolerance of polli-
nation networks to species extinctions. Proc R Soc London Ser B
biol Sci 271: 2605 LP–2602611.
Neutel, A.-M., Heesterbeek, J. A. P., and de Ruiter, P. C. (2002). Stability
in real food webs: Weak links in long loops. Science 296(80): 1120
LP–1121123.
Norton, B. G. (2000). Population and consumption: Environmental prob-
lems as problems of scale. Ethics Environ 5: 23–45.
O’Connell, J. F., and Hawkes, K. (1984). Food choice and foraging sites
among the Alyawara. J Anthropol Res 40: 504–535.
Orians G, Pearson N (1979) On the theory of central place foraging. In:
Horn D, Mitchell R, Stairs G (eds) Advances in Ecological Systems.
pp 154–177
Pimm, S. L. (1979). The structure of food webs. Theor Popul Biol 16:
144–158. https://doi.org/10.1016/0040-5809(79)90010-8.
Post, W. M., and Pimm, S. L. (1983). Community assembly and food web
stability. Math Biosci 64: 169–192. https://doi.org/10.1016/0025-
5564(83)90002-0.
Ready, E., and Power, E. A. (2017). Why wage earners hunt: Food shar-
ing, social structure, and influence in an Arctic mixed economy.
Curr Anthropol 59: 74–97. https://doi.org/10.1086/696018.
Rezende, E. L., Lavabre, J. E., Guimarães, P. R., et al (2007). Non-
random coextinctions in phylogenetically structured mutualistic net-
works. Nature 448: 925.
Romanuk, T. N., Zhou, Y., Brose, U., et al (2009). Predicting invasion
success in complex ecological networks. Philos trans R Soc B biol
Sci 364: 1743 LP–1741754.
Romanuk TN, Zhou Y, Valdovinos FS, Martinez ND (2017) Robustness
trade-offs in model food webs: Invasion probability decreases while
invasion consequences increase with Connectance, 1st edn. Elsevier
Ltd.
Scelza, B. A. (2012). Food scarcity, not economic constraint limits con-
sumption in a rural aboriginal community. Aust J Rural Health 20:
108–112.
Scelza, B. A., Bird, D. W., and Bliege Bird, R. (2014). Bush Tucker, shop
Tucker: Production, consumption, and diet at an aboriginal outsta-
tion. Ecol Food Nutr 53: 98–117. https://doi.org/10.1080/03670244.
2013.772513.
Schoener, T. W. (1979). Generality of the size-distance relation in models
of optimal feeding. Am Nat 114: 902–914. https://doi.org/10.1086/
283537.
Smith-Ramesh, L. M., Moore, A. C., and Schmitz, O. J. (2017). Global
synthesis suggests that food web connectance correlates to invasion
resistance. Glob Chang Biol 23: 465–473. https://doi.org/10.1111/
gcb.13460.
Solé, R. V., and Montoya, M. (2001). Complexity and fragility in ecolog-
ical networks. Proc R Soc London Ser B biol Sci 268: 2039 LP–
2032045.
Srinivasan, U. T., Dunne, J. A., Harte, J., and Martinez, N. D. (2007).
Response of complex food webs to realistic extinction sequences.
Ecology 88: 671–682. https://doi.org/10.1890/06-0971.
Sutherland, D. R., Glen, A. S., and de Tores, P. J. (2011). Could control-
ling mammalian carnivores lead to mesopredator release of carniv-
orous reptiles? Proc R Soc B Biol Sci 278: 641–648. https://doi.org/
10.1098/rspb.2010.2103.
Svanbäck, R., and Bolnick, D. I. (2007). Intraspecific competition drives
increased resource use diversity within a natural population. Proc R
Soc B Biol Sci 274: 839–844. https://doi.org/10.1098/rspb.2006.
0198.
Terborgh, J., Lopez, L., Nuñez, P., et al (2001). Ecological meltdown in
predator-free Forest fragments. Science 294(80): 1923 LP–
1921926.
Thompson, R. M., Brose, U., Dunne, J. A., et al (2012). Food webs:
Reconciling the structure and function of biodiversity. Trends Ecol
Evol 27: 689–697. https://doi.org/10.1016/j.tree.2012.08.005.
Tylianakis, J. M., Laliberté, E., Nielsen, A., and Bascompte, J. (2010).
Conservation of species interaction networks. Biol Conserv 143:
2270–2279. https://doi.org/10.1016/j.biocon.2009.12.004.
Walsh, F. J. (1987). The influence of the spatial and temporal distribution
of plant food resources on traditional Martujarra subsistence strate-
gies. Aust Archaeol: 88–101.
Walsh, F. (1990). An ecological study of traditional aboriginal use of
‘Bcountry^’: Martu in the great and little Sandy deserts, Western
Australia. Proc Ecol Soc Aust 16: 23–37.
Walsh F (2008) To hunt and to hold: Martu aboriginal people’susesand
knowledge of their country, with implications for co-management in
Karlamilyi(Rudall River) National Park and thegreat Sandy Desert,
Western Australia. The University of Western Australia
Walsh, F., and Douglas, J. (2011). No bush foods without people: The
essential human dimension to the sustainability of trade in native
plant products from desert Australia. Rangel J 33: 395–416.
Hum Ecol
Author's personal copy

Walsh, F. J., Dobson, P. V., and Douglas, J. C. (2013). Anpernirrentye: A
framework for enhanced application of indigenous ecological
knowledge in natural resource management. Ecol Soc 18. https://
doi.org/10.5751/ES-05501-180318.
Williams RJ (2010) Network3D software. In: Microsoft Research.
Cambridge, U. K.
Williams, R. J., Berlow, E. L., Dunne, J. A., et al (2002). Two degrees of
separation in complex food webs. Proc Natl Acad Sci 99: 12913–
12916. https://doi.org/10.1073/pnas.192448799.
Winterhalder, B. (1981). Foraging strategies in the boreal forest: An anal-
ysis of Cree hunting and gathering. In Winterhalder, B., and Smith,
E. A. (eds.), Hunter gatherer foraging strategies: Ethnographic and
archeological analyses. University of Chicago Press, Chicago,
Illinois, pp. 66–98.
Woinarski J, Legge S, Fitzsimons J, et al (2011) The disappearing mam-
mal fauna of northern Australia: Context, cause, and response
Woinarski, J.C. Z., Burbidge, A. A., and Harrison, P. L. (2015). Ongoing
unraveling of a continental fauna: Decline and extinction of
Australian mammals since European settlement. Proc Natl Acad
Sci 112: 4531 LP–4534540.
Yoon I, Williams R, LevineE, et al (2004) Webs on the web (WOW): 3D
visualization of ecological networks on the WWW for collaborative
research and education. Pp 124–132
Zeanah, D. W., Codding, B. F., Bird, D. W., et al (2015). Diesel and
damper: Changes in seed use and mobility patterns following con-
tact amongst the Martu of Western Australia. J Anthropol Archaeol
39: 51–62. https://doi.org/10.1016/j.jaa.2015.02.002.
Zeanah, D. W., Codding, B. F., Bird, R. B., and Bird, D. W. (2017).
Mosaics offire and water: The co-emergence ofanthropogenic land-
scapes and intensive seed exploitation in the Australian arid zone.
Aust Archaeol 83: 2–19. https://doi.org/10.1080/03122417.2017.
1359876.
Zhu, Y., Yan, X., and Moore, C. (2014). Oriented and degree-generated
block models: Generating and inferring communities with inhomo-
geneous degree distributions. J Complex Networks 2: 1–18. https://
doi.org/10.1093/comnet/cnt011.
Ziembicki M, Woinarski J, Mackey B (2013) Evaluating the status of
species using Indigenous knowledge: Novel evidence for major na-
tive mammal declines in northern Australia
Ziembicki,M.R.,Woinarski,J.C.Z.,Webb,J.K.,etal(2015).
Stemming the tide: Progress towards resolving the causes of decline
and implementing management responses for the disappearing
mammal fauna of northern Australia. Therya 6: 169–226. https://
doi.org/10.12933/therya-15-236.
Hum Ecol
Author's personal copy