ArticlePDF Available

Abstract and Figures

1. Trophic cascade theory predicts that apex predators structure ecosystems by regulating mesopredator and herbivore abundance and behaviour. Studies on trophic cascades have typically focused on short linear chains of species interactions. A framework that integrates more realistic and complex interactions is needed to make broader predictions on ecosystem structuring. 2. Network analysis is used to study food webs and other types of species interaction networks. These often comprise large numbers of species but rarely account for multiple interaction types and strengths. Here we develop an intermediate complexity theoretical framework that allows specification of multiple interaction types and strengths for the study of trophic cascades. This ecological network is designed to suit data typically derived from field-based studies. The trophic cascade network contains fewer nodes than food webs, but provides semi-weighted directional links that enable different types of interactions to be included in a single model. 3. We use this trophic cascade network model to explore how an apex predator shapes ecosystem structure in an Australian arid ecosystem. We compared two networks that contrasted in the dominance of an apex predator, the dingo (Canis dingo), using published results ranking the direction and strength of key interactions. Nodes and links interacted dynamically to shape these networks. We examined how changes to an apex predator population affects ecosystem structure through their direct and indirect influences on different components of this ecological community. 4. Under strong apex predator influence, the network structure was denser and more complex, even, and top-down driven; and dingo predation and soil commensalism formed denser interactive modules. Under weak apex predator influence (e.g. reflecting predator control) the resulting network structure was frayed, with mesopredator predation and grazing forming modules. 5. Our study demonstrates that networks of intermediate complexity can provide a powerful tool for elucidating potential ecosystem-wide effects of apex predators, and predicting the consequences of management interventions such as predator control. Integrating trophic cascades, with their array of complex interactions, with the three-dimensional structure of ecological networks, has the potential to reveal ‘ecological architecture’ that neither captures on its own.
Content may be subject to copyright.
Trophic cascades in 3D: network analysis reveals how
apex predators structure ecosystems
Arian D. Wallach
1
*, Anthony H. Dekker
2
, Miguel Lurgi
3,4
, Jose M. Montoya
3
,DamienA.
Fordham
4
andEuanG.Ritchie
5
1
Centre for Compassionate Conservation, School of Life Sciences, University of Technology Sydney, PO Box 123 Broadway,
Ultimo NSW 2007, Australia;
2
Federation University (Ballarat), Mt Helen, PO Box 663, Ballarat Vic. 3353, Australia;
3
Ecological
Networks and Global Change Group, Theoretical and Experimental Ecology Station, CNRS and Paul Sabatier University,
Moulis 09200, France;
4
The Environment Institute & School of Biological Sciences, University of Adelaide, Adelaide SA 5005,
Australia; and
5
School of Life and Environmental Sciences, Centre for Integrative Ecology (Burwood Campus), Deakin
University, 221 Burwood Highway, Burwood, Vic. 3125, Australia
Summary
1. Trophic cascade theory predicts that apex predators structure ecosystems by regulating mesopredator and
herbivore abundance and behaviour. Studies on trophic cascades have typically focused on short linear chains of
species interactions. A framework that integrates more realistic and complex interactions is needed to make
broader predictions on ecosystem structuring.
2. Network analysis is used to study food webs and other types of species interaction networks. These often com-
prise large numbers of species but rarely account for multiple interaction types and strengths. Here, we develop
an intermediate complexity theoretical framework that allows specification of multiple interaction types and
strengths for the study of trophic cascades. This ecological network is designed to suit data typically derived from
field-based studies. The trophic cascade network contains fewer nodes than food webs, but provides semi-
weighted directional links that enable different types of interactions to be included in a single model.
3. We use this trophic cascade network model to explore how an apex predator shapes ecosystem structure in an
Australian arid ecosystem. We compared two networks that contrasted in the dominance of an apex predator,
the dingo (Canis dingo), using published results ranking the direction and strength of key interactions. Nodes and
links interacted dynamically to shape these networks. We examined how changes to an apex predator population
affect ecosystem structure through their direct and indirect influences on different components of this ecological
community.
4. Under strong apex predator influence, the network structure was denser and more complex, even and top-
down driven; and dingo predation and soil commensalism formed denser interactive modules. Under weak apex
predator influence (e.g. reflecting predator control), the resulting network structure was frayed, with mesopreda-
tor predation and grazing forming modules.
5. Our study demonstrates that networks of intermediate complexity can provide a powerful tool for elucidating
potential ecosystem-wide effects of apex predators and predicting the consequences of management interventions
such as predator control. Integrating trophic cascades, with their array of complex interactions, with the three-
dimensional structure of ecological networks, has the potential to reveal ‘ecological architecture’ that neither cap-
tures on its own.
Key-words: bioturbation, dingo, ecosystem structure, food webs, mutualism, predation, top-down
regulation
Introduction
The role of apex predators as ecosystem regulators is now
firmly embedded in ecological theory, suggesting that the
world is green and biologically diverse in large part because
predators suppress herbivore densities (Hairston, Smith &
Slobodkin 1960; Estes et al. 2011; Ripple et al. 2014). Studies
from across the globe show that apex predators limit the
abundance and modify the behaviour of their prey and smal-
ler mesopredators, suppressing grazing and predation pres-
sure, and enhancing biodiversity and productivity (Ritchie &
Johnson 2009; Ritchie et al. 2012). This top-down forcing
cascades throughout ecosystems influencing a broad range of
processes, both biotic and abiotic, including species abun-
dances and richness, animal behaviour, disease dynamics, car-
bon sequestration and stream morphology (Estes et al. 2011;
Ripple et al. 2014; Atwood et al. 2015). The rise and fall of
apex predators not only affects the composition of species
*Correspondence author. E-mail: arian.wallach@uts.edu.au
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society
Methods in Ecology and Evolution 2016 doi: 10.1111/2041-210X.12663
within ecological communities therefore, but also ecosystem
functioning (Estes et al. 2011; Ripple et al. 2014; Standish
et al. 2014). For example, wolves (Canis lupus) provide critical
resource subsidies to scavenging species during warm months,
thus enhancing their resilience to shortening winters due to
global warming (Wilmers & Getz 2005). Similarly, dingoes
(C. dingo) stabilize herbivore prey densities by dampening
their population responses to rainfall in arid environments,
thereby enabling plant biomass to accumulate during brief
wet seasons (Letnic & Crowther 2013).
Trophic cascades are typically studied as relatively short and
hierarchical chains of interactions, tested for relative strength
and direction (e.g. predator ?
herbivore ?
vegetation)
(Bascompte & Stouffer 2009; Ritchie & Johnson 2009).
Trophic cascade theory, however, aims to explain much
broader patterns in nature and is therefore well placed to be
studied in an ecological network context (Montoya, Pimm &
Sole 2006; Bascompte 2009). Ecological network analysis can
be used to explore questions pertaining to community struc-
ture and dynamics, and to provide a platform for identifying
features that maintain and enhance biodiversity (Montoya,
Pimm & Sole 2006; Bascompte 2009; Thompson et al. 2012).
For example, networks have been used to identify keystone
species, elements and trophic structures that confer resistance
to different types of perturbations, and to investigate the influ-
ence of adding or removing species from ecosystems (Mon-
toya, Pimm & Sole 2006; Bascompte 2009; S
aterberg, Sellman
& Ebenman 2013). Furthermore, ecological networks provide
a powerful tool for exploring the interconnectivity of nature
and for predicting the robustness or fragility of ecosystem
states (Montoya, Pimm & Sole 2006; Pascual & Dunne 2006).
They constitute our main tool for understanding the relation-
ship between diversity and stability in natural communities
(Allesina & Tang 2012).
Ecological network studies have traditionally focused on
feeding interactions and mutualisms (Ings et al. 2009; K
efi
et al. 2012), but trophic cascade studies often include other
types of interactions (e.g. interspecific killing, risk effects and
competition) that vary in their strength (Creel & Christianson
2008; Ritchie & Johnson 2009). Large predators often hunt a
variety of species, but their population level effect is usually
restricted to only some of theirprey.Forexample,dingoes
prey on a wide range of animals, from very small (<1kg)to
very large (>100 kg), but they primarily suppress populations
of medium to large animals (Letnic, Ritchie & Dickman
2012). Thus, the indirect effect of a large predator on a prey
species can be positive if it suppresses another predator that
exerts a stronger predation force on that prey (Letnic, Ritchie
& Dickman 2012). Network analyses of trophic cascade stud-
ies are therefore well suited to an intermediate complexity
approach that incorporates the strength and type of trophic
interactions derived from well-studied relationships.
Understanding the importance of predator loss (Ripple
et al. 2014) and reestablishment (Chapron et al. 2014) is of
widespread theoretical and management interest, due to its rel-
evance for actions such as limiting and recovering wildlife pop-
ulations (Wallach et al. 2010; Ritchie et al. 2012; Newsome
et al. 2015). Integrating trophic cascades, with their array of
complex interactions, with the three-dimensional structure of
ecological networks, has the potential to reveal ‘ecological
architecture’ that neither captures on its own. The first aim of
our study was to develop a network analysis method suitable
for trophic cascade field studies, which incorporates different
types, and varying strengths, of interactions into a single
model. Our second aim was to examine and demonstrate the
types of insights that arise from networks on the ecological role
of apex predators. To achieve this, we developed a network
model of well-studied trophic interactions including both sup-
pressive and commensal interactions. We constructed the eco-
logical network from several highly interactive species of the
Australian arid zone (Glen & Dickman 2005; Dickman et al.
2014) and examined how ecosystem structure may respond to
a functionally dominant or weakened dingo population.
Australia’s apex predator, the dingo, plays a keystone role
in enhancing biodiversity by limiting herbivore prey (e.g. kan-
garoos, Macropus spp.) and mesopredators (e.g. red foxes,
Vulpes vulpes) (Letnic, Ritchie & Dickman 2012). Widespread
persecution of dingoes is now understood to be a leading cause
of a series of mammal extinctions across the continent (John-
son 2006), many of which played key ecosystem functions
(Fleming et al. 2014). Medium-sized (critical weight range)
mammals (355500 g) in arid environments have been particu-
larly vulnerable to predation by mesopredators (Johnson &
Isaac 2009). Many of Australia’s digging mammals fall within
this critical weight range, and consequently their bioturbation
(soil disturbance) effects have declined. This ecological func-
tion enhances soil properties, such as turnover, organic matter
and water infiltration, which promotes plants and provides
habitat for other organisms (Fleming et al. 2014). Thus, sup-
pressive feeding interactions by dingoes can cascade to influ-
ence mutualisms driven by other species.
We investigated the top-down effects of the dingo on ecosys-
tem structure and function by comparing two scenarios: in the
first, the dingo population is intact, and in the second, the
dingo population is suppressed. Our model system predicts
that suppressing the ecologicalroleofdingoescanprovoke
structural changes to ecosystems resulting in shifts between
alternative ecosystem states.
Materials and methods
Ecological networks consist of ecosystem units (e.g. species) repre-
sented as nodes that are connected through ecological relationships
(e.g. trophic) represented as links.Bothnodesandlinkscanvaryin
their weight,wherenode weights can represent a species’ population
size, biomass or ecological effect, and link weights can represent the
strength (e.g. effect size) and type (e.g. predation) of interactions. For
clarity, throughout this paper, species and elements are capitalized
when referred to as nodes in the network (e.g. ‘dingo’ refers to the
species and ‘Dingo’ refers to the node).
NETWORK COMPONENTS
We constructed an ecological network comprising nine nodes (Table 1)
chosen to represent well-studied highly interactive species and elements
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society, Methods in Ecology and Evolution
2A. D. Wallach et al.
of the Australian arid ecosystem (Glen & Dickman 2005; Dickman
et al. 2014). We focused on the arid zone, which encompasses about
70% of the continent, because most extinctions and range contractions
and most trophic cascade studies have occurred in this region (John-
son & Isaac 2009; Letnic, Ritchie & Dickman 2012). We incorporated
both suppressive interactions predation and herbivory and mutu-
alisms bioturbation and the effects of plants on soil.
We chose the dingo to represent an apex predator and focused the
network analysis on how changes to this one species trigger shifts in
ecosystem structure. The red fox and wild cat (Felis catus)were
included in the network to represent highly interactive mesopredators.
Herbivores were represented by rabbits (Oryctolagus cuniculus)and
kangaroos. The greater bilby (Macrotis lagotis) was chosen to represent
a non-herbivorous digging mammal that is threatened by mesopreda-
tor predation. Bilbies, rabbits and small mammals were all included as
ecosystem engineers through their bioturbation effects. Small mam-
mals, vegetation and soil were included as functional groups and
ecosystem properties.
Trophic cascade studies traditionally focus on small sets of interac-
tions,and we broughtthree studiestogether to develop our model. Link
weights between the Dingo, Fox, Cat, Kangaroo, Rabbit, Small mam-
mal and Vegetation nodes were assigned from the results of generalized
linear models and principle component analyses reported in a trophic
cascade study by Wallach et al. (2010). The network was expanded to
include two additional nodes: Bilby and Soil to illustrate how studies
can be combined to provide predictive tools to assess how the recovery
or extirpation of an apex predator can affect ecosystem functions. Link
weights generated from the Dingo, Fox and Cats nodes to the Bilby
node were assigned from the generalized linear model reported in
Southgate et al. (2007), and the effects of mammalian bioturbation by
Rabbit, Bilby and Small mammal nodes on Soil were ranked from
measurements conducted by James, Eldridge & Hill (2009). All three
studies were conducted in the arid zone and together, when unified into
an ecological network, provided a predictive m odel of how the recovery
or suppression of dingoes may affect eco system function.
INCORPORATING INTERACTION STRENGTHS INTO A
NETWORK MODEL
Weappliedasetofrulestotranslatethe results from the selected studies
on interaction strengths into link weights on a discrete scale ranging
from 3to+3, to represent strongly suppressive to strongly mutualistic
interactions (Table S1, Supporting Information). For example,
Dingo?Fox was assigned a link weight of 3, while the Dingo?Cat
link was only ranked 2, because the models in the focal study (Wal-
lach et al. 2010) show a stronger (94) suppressive effect of dingoes on
foxes than on cats (Table S2). This qualitative method for inferring
interaction strengths enables different types of interactions (e.g. preda-
tion and bioturbation) to be included in a single model.
To simplify the analysis, each interaction type was assigned a fixed
negative or positive value. For example, herbivory was always assigned
a negative link value even though it can also be mutualistic (e.g. herbi-
vores also promote the growth and reproduction of plants). Links rep-
resented direct interactions between pairs of nodes (e.g. Dingo?
Kangaroo), while indirect interactions (e.g. trophic cascades, Dingo?
Vegetation) were calculated from the closest set of links between dis-
connected nodes. Links were assigned a single direction from the ‘af-
fecting’ to ‘affected’ nodes (e.g. the influence of a predator on a prey
was included, but not vice versa). The three studies yielded 20 paired
interactions varying in weight and direction (Table 2).
MODELLING TROPHIC CASCADES AS A NETWORK
The set of nine nodes and their 20 paired links formed the network
structure. These were used to model how changes to the apex predator
node trigger changes to the network structure. Node weights were
assigned discrete values ranging from 1 to 3, representing a species’ or
element’s (weak to strong) interactive strength within the network.
Two versions of the network were derived representing two ecological
states (ES) based on the functional condition of the apex predator pop-
ulation. In ES1, the weight of the Dingo node was ranked high
Table 1. Elements used to construct the network
Functional role
Representative
species/element
Apex predator Dingo
Mesopredator Fox
Mesopredator Cat
Large herbivore Kangaroo
Medium herbivore and ecological engineer
(bioturbation agent)
Rabbit
Small mammal Small mammal
Medium insectivore and ecological engineer
(bioturbation agent)
Bilby
Primary productivity Vegetation
Soil Soil
Table 2. Maximum link weights assigned based on key literature assessing ecological interaction strengths. A nil interaction was assigned where no
significant interaction was detected in the studies, even if such interactionsdo exist in nature. Node Ais affecting Node Bbut not vice versa. For refer-
ence details, see Table S2. Cell colours vary from red to green highlighting the corresponding values ranging from 3to+3
A
B
Fox Cat Kangaroo Rabbit Small mammals Bilby Vegetation Soil
Dingo 3232110 0
Fox 10 1130 0
Cat 0 1220 0
Kangaroo 0 0 0 30
Rabbit 003+2
Small mammals 01+1
Bilby 0+2
Vegetation +3
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society, Methods in Ecology and Evolution
Trophic cascades in 3D 3
(Dingo =3), representing a condition in which the dingo population is
present without restrictions. In ES2, the Dingo node weight was ranked
low (Dingo =1), to model a situation in which the apex predator is
functionally absent or suppressed (e.g. subjected to lethal control). The
effect of changing the weight of the Dingo node ‘cascaded’ throughout
the network through a set of ‘game rules’ that determined the relation-
ship between node and link weights (Box 1).
Let node Arepresents the affecting species/element (e.g. predator)
and node Bthe affected species/element (e.g. prey) in each pair. The
node weights are denoted as Node A/B=X,whereX=1, 2 or 3. The
link weights are denoted A?
X
B, and the value of Xranges discretely
from 3to+3.ThenodeweightofAcombined with the link weight
determined the node weight of B. The three key reference studies pro-
vided the maximum link weights when the node weight of Awas maxi-
mal (denoted A
max
) (Table S2). If the node weight of A declined, so did
its link weight and thus its overall effect in the network. The node
weight of Bwas then determined by the adjusted link weight. For sim-
plicity, the weight of node Bwas defined by the strongest interaction
and was not cumulative.
Thus, suppressive interactions resulted in weaker nodes and
weaker links, while mutualism interactions increased them. For
example, a suppressive predatorprey interaction reduces the node
weight of the prey and also the link weight generated by the prey.
Thus, links between nodes that are connected via a trophic (feeding)
interaction could be severed if the node weight and its associated link
weight were sufficiently weakened. This represents interactions in nat-
ure in which feeding interactions do not result in discernible popula-
tion level effects.
Assigning a maximal weight for the Dingo node (Dingo =3) in ES1
and a minimal weight (Dingo =1) in ES2, changed the node weights,
adjusted the link weights and the number of links. Some links severed
when the effect size became too low, leaving a total of 15 links in ES1
and 12 links in ES2 (Table S3).
NETWORK ANALYSIS
The adjusted node and link weights forming the two networks
(Table S3) were analysed for four main properties: distance, quantita-
tive degree, centrality and connectance.
Distance is a weighted measure of how close a given node is to
another and represents its relative influence on it. Unlike link weights,
this variable shows the influence of one node on another regardless of
whether there are direct interactions between them. Distance is calcu-
lated using the units of link weights between pairs of nodes, and if the
nodes are not linked, the distance used is calculated as the shortest path
between them via other nodes (high link weights reduces the distance
between nodes). We compared the average, standard deviation (SD)
and coefficient of variance (CV) of distances, and identified modules of
Box 1. The dynamic relation betweennodeweightandlinkweight
Node and link weights interact dynamically to shape the network following a set of ‘game rules’. The published studies determined the link
weights when the node weights are maximal (Table 2). When the weight of node Ais reduced, so is its effect in thenetworkanditslinkweightis
also reduced (Table I). This adjusted link weight then determines the node weight of B(Table II). The relation between node and link weights is
illustrated in Fig. I.
Table I. Maximum link weight (A
max
?
X
B) and the node weight of A(A:X) determine the adjusted link weight (A-X-B). As A:Xdeclines,
link weight declines and in some cases the link severs
A=X
A
max
?
X
B
321123
A=3A?
3
BA?
2
BA?
1
BA?
+1
BA?
+2
BA?
+3
B
A=2A?
2
BA?
1
B// A?
+1
BA?
+2
B
A=1A?
1
B/// / A?
+1
B
Table II. The adjusted link weight (A?
X
B) determines the node weight of B(B=X)
Link weight Node weight
A?
3
BB=1
A?
2
BB=2
A?
1
BB=3
A?
+1
BB=1
A?
+2
BB=2
A?
+3
BB=3
Fig. I. Illustration of how link and node weights cascade through the network.
Dingo:3
Fox:1
Link and node weights cascade through the network
Bilby:3
Veg: 3
Soil:3
3
+3
Cat:2
2
+2
Kangaroo:1
–3
1
1
1
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society, Methods in Ecology and Evolution
4A. D. Wallach et al.
higher density (lowest distances). We used a paired t-test (after verifying
normal distribution, using a quantilequantile plot) to compare dis-
tances between pairs of nodes in ES1 and ES2, and we identified mod-
ules (denser regions in the network) of node pairs with distances <1and
which differed by 92ormorebetweenES1andES2.
Weighted degree represents the local importance of each node by its
weighted connectivity within the network and is calculated by summing
the absolute values of all the link weight values connected to that node.
We compared the average (with a Paired t-test), SD and CV of node
weights between the two networks.
Centrality is a measure that quantifies how close a given node is to
every other node in the network. It is a measure commonly used to
determine how important a node is globally based on its role as a con-
nector between nodes. It is calculated as the average of the reciprocals
of the network distances to each node as:
CvðxÞ¼ 1
n1X
yx
1
dðx;yÞ
!
;
where C
v
(x) is the centrality of node x,nis the number of nodes in the
network, and d(x,y) is the network distance between nodes xand y(for
directly linked nodes, this will simply be the reciprocal of the link
(a)
(b)
Fig. 1. Network structures of the two ecosystem states (ES) ES1 and
ES2. In ES1, the Dingo node was assigned high weight score (a) and in
ES2 a low weight score (b). The transition between the two states is
shown in Video S1 (c). The volume of each ball indicates node weight,
the thickness of lines represents link weight, and the length of lines
denotes link distance. Colours range from red (low centrality score) to
green (high centrality score). Centrality and link distance are scaled
independently within each diagram.
Table 3. Network structure of the two ecosystem states (ES) featuring
the properties distance (a), degree (b) and centrality (c). Cells high-
lighted in green and blue have lower distances (>2) (a), and highest
degree (b) and centrality (c) scores, for ES1 and ES2, respectively
(a)
Distance ES1 ES2
DingoFox 031
DingoCat 0518
DingoKangaroo 031
DingoRabbit 0517
DingoBilby 1 13
DingoSmall mammal 1 2
DingoVegetation 1 13
DingoSoil 1523
FoxCat 0808
FoxKangaroo 0717
FoxRabbit 081
FoxBilby 1 03
FoxSmall mammal 131
FoxVegetation 1313
FoxSoil 1523
CatKangaroo 0817
CatRabbit 1 1
CatBilby 1 05
CatSmall mammal 1 05
CatVegetation 1513
CatSoil 1523
KangarooRabbit 0807
KangarooBilby 132
KangarooSmall mammal 1322
KangarooVegetation 1 03
KangarooSoil 1513
RabbitBilby 1513
RabbitSmall mammal 1515
RabbitVegetation 0503
RabbitSoil 1 13
BilbySmall mammal 2 1
BilbyVegetation 1 17
BilbySoil 0527
Small mammalVegetation 1 18
Small mammalSoil 1528
Vegetation-Soil 051
Average 103 139
SD 041 066
CV (%) 3939 4778
Accumulated 37250
(b)
Degree ES1 ES2
Dingo 12 2
Fox 4 7
Cat 4 6
Kangaroo 4 4
Rabbit 5 5
Bilby 5 5
Small mammal 3 3
Vegetation 6 7
Soil 5 1
Average 5344
SD 2621
CV (%) 496478
Accumulated 48 40
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society, Methods in Ecology and Evolution
Trophic cascades in 3D 5
weight). This definition of centrality, which differs from the more gen-
eral usage (the reciprocal of the average distance), is more suitable for
ecological network analysis because it remains well defined even if
removal of a species results in disconnection of the network, causing
some of the d(x,y) to become infinite (Dekker 2005). We compared the
average (with a paired t-test), SD and CV of centrality values between
the two networks.
Connectance assesses the level of complexity of the network, by
quantifying the density of interactions through the fraction of realized
(out of the possible) links in the network:
C¼L
nðn1Þ;
where Cis the network’s connectance, Lis the number of links, and nis
the number of nodes (Pimm, Lawton & Cohen 1991).
Results
The node weights and adjusted link weights of ES1 and ES2
structured two distinct networks (Fig. 1). When the Dingo
node weight was high (ES1), the network was denser, with
26% lower average distances between nodes, compared to the
network in which the Dingo node was weakened (ES2)
(t=31, d.f. =35, P<001). ES1 was also more evenly
shaped, with a lower CV of distances (Table 3a), and was more
complex (C=018) than ES2 (C=013).
In the ES1 network, the Dingo was the most central and
interconnected (degree score) node (Table 3b,c). In contrast,
in ES2 the Vegetation and Fox nodes had the highest degree
scores, and Vegetation was most central in the network
(Table 3b,c). The average degree and centrality scores were
1820% higher in ES1 compared to ES2, although these differ-
ences were not statistically significant. The degree and central-
ity scores of the Dingo and Soil nodes declined considerably
when the Dingo node was weakened (Table 3b,c).
Distances between some node pairs differed substantially
between ES1 and ES2 (Table 3a). In ES1, the Dingo node was
at least three times closer to the Kangaroo, Fox, Cat and Rab-
bit nodes, and the Bilby node was over five times closer to Soil,
compared to ES2. In ES2, the Fox node was three times closer
to Bilby, and Kangaroo was three times closer to Vegetation,
compared to ES1 (Table 3a).
These changes in distances formed internal modules of
higher density (low distances). ES1 formed one module com-
prising of dingo predation interactions (DingoCat/Fox/Kan-
garoo/Rabbit) and a second module of soil commensals
(Vegetation/BilbySoil). ES2 formed a module of mesopreda-
tor predation (e.g. Fox/CatBilby/Small mammal) and of
grazing (KangarooVegetation) (Table 3a). In both ES1 and
ES2, FoxCat/Rabbit and RabbitVegetation remained simi-
larly close.
Discussion
Network analysis can bring new insights into trophic cascade
studies, complementing existing analysis tools. Using a net-
work model of intermediate complexity, we showed how the
direct effects of an apex predator on its prey influence funda-
mental network properties. We detected four main structural
differences between the two modelled ecosystem states: density,
complexity, evenness and top-down forcing. When the Dingo
node was assigned a high score (ES1), the resulting network
structure was denser, more even and complex and top-down
forces dominated. By contrast, when the Dingo node was sup-
pressed (ES2), the network structure was frayed and top-down
forces were weakened. Our network analysis therefore suggests
that the loss of apex predators leads to the ‘unravelling’ of
ecosystems, consistent with theory (Estes et al. 2011).
In Australia, and globally, the decline of apex predators is
often associated with increasing mesopredator predation and
grazing pressure, which can shift ecosystems to alternative
states (Wolf, Cooper & Hobbs 2007; Wallach et al. 2010; Rip-
ple et al. 2014). Our network analysis revealed how changes in
the status of the apex predator alter direct and indirect interac-
tions between other species, forming contrasting ecosystem
states. ES1 had modules around apex predator predation and
soil mutualisms, and the Dingo node was highly intercon-
nected and central. In contrast, ES2 had modules around
mesopredator predation and grazing, the Vegetation and Fox
nodes were the most interconnected, and Vegetation was cen-
tral.Ourmodelthereforepredicts that increasing top-down
forces by allowing dingoes to recover from lethal control is
likely to benefit animals vulnerable to mesopredator predation
(e.g. foxes?bilbies) and promote their ecological function
(e.g. bioturbation).
This suggests more broadly that top-down regulated ecosys-
tems can be conducive to a range of mutualism interactions by
other species. For example, beavers (Castor canadensis)drive
mutualisms with other plants and animals by damming creeks.
The eradication of wolves from Yellowstone National Park,
North America, increased elk (Cervus elaphus)browsingto
levels that excluded beavers, which shifted the stream habitat
from ponds and floodplains supporting structurally complex
vegetation to an alternative state that is channelled, eroded
and surrounded by open grassland (Wolf, Cooper & Hobbs
2007). Similarly, predatory fish promote mutualisms between
insect pollinators and plants, by feeding on the aquatic larval
stage of predatory dragonfly (Knight et al. 2005). These cas-
cades can be complex, however: wolves can also suppress
(c)
Centrality ES1 ES2
Dingo 171 07
Fox 126 112
Cat 109 107
Kangaroo 126 105
Rabbit 122 116
Bilby 099 112
Small mammal 079 082
Vegetation 118 13
Soil 104 056
Average 117 099
SD 025 024
CV (%) 2157 2442
Accumulated 10589
Table 3. (continued)
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society, Methods in Ecology and Evolution
6A. D. Wallach et al.
beavers (Potvin et al. 1992; Rosell & Sanda 2006), and preda-
tors of mutualists can also have negative effects on plants (e.g.
birds eating pollinating insects) (Knight et al. 2006).
We developed the current network from interaction
strengths ranked according to single analyses, from a set of
chosen studies, and it is likely that other data sets will yield dif-
fering results. The consistency of outcomes arising from net-
work analyses is probably similar to that of other models,
which are affected by natural and methodological variations
between studies. Overall, we expect that our results are robust
because the ecological effects of dingoes are typically consistent
(Letnic, Ritchie & Dickman 2012). Studies conducted in
deserts and forests have yielded strikingly similar results (Col-
man et al. 2014). Some variation between studies does exist,
however. For example, we ranked the effect of dingoes on rab-
bits as quite strongly negative (following the results of Wallach
et al. 2010), while other studies have reported positive interac-
tions (Letnic, Ritchie & Dickman 2012).
A more comprehensive network analysis of trophic cascades
would involve not only a larger number of nodes, but also
dynamic bidirectional links. Here, for example, we focused on
the top-down effect of the predator on the prey, excluding the
bottom-up (resource) effects of prey on predators. These two-
way interactions are important for investigating dynamic pro-
cesses such as feedback loops (e.g. between plants and soil).
Dynamic interactions also exist within species. For example,
the mutualistic relationships within plant communities can
trigger positive feedback loops that promote plant growth
(McAlpine et al. 2009), and carnivore social behaviour can
suppress population growth (Wallach et al. 2015). Future
studies could also consider more nuanced interactions. We
ranked trophic interactions as purely suppressive, even though
herbivores also benefit plants, and we ranked animalsoil inter-
actions as purely commensal, even though animals can also
degrade soil.
Our study provides a proof of concept for the use of network
analysis in the study of trophic cascades and highlights the ben-
efits of adopting an intermediate complexity approach for
analysis of field-based research. The approach extends trophic
cascades from linear interactions, to system-level processes.
The analysis demonstrates how networks could incorporate
interactions that drive population dynamics, since not all feed-
ing interactions drive populations. Mesoscale studies of eco-
logical networks can reveal patterns in community assembly
that are hard to study on large ecological networks and are not
detectable at small (module) scales (Bascompte & Stouffer
2009). Finally, our study also provides a demonstration of how
disparate field studies, with varying types of quantitative infor-
mation, can be assembled into a network. For example, we
extended a trophic cascade study (Wallach et al. 2010) by two
nodes Bilby and Soil (Southgate et al. 2007; James, Eldridge
& Hill 2009) to generate testable predictions on how the
recovery of dingoes could increase mutualism interactions by a
threatened ecosystem engineer [dingo?
mesopredator?
bilby?
+
soil]. This is important because few studies are able
to provide quantitative information on many nodes and links
on their own.
Networks provide a helpful tool for integrating multiple
interaction types within an ecosystem. They allow, for exam-
ple, combining predatorprey interactions with ecosystem
engineering (e.g. bioturbation) effects, as we have shown here.
Such complexities constitute one of the biggest challenges in
network ecology, affecting the structure, dynamics and func-
tioning of communities (Ings et al. 2009; K
efi et al. 2012). Our
method (or an adaptation thereof) can be applied to the analy-
sis of primary data sets, systematic reviews and theoretical
studies, to help investigate ‘big picture’ questions and model
scenarios that can be difficult to implement in the field.
Network-based ecological models can generate testable
hypotheses on the consequences of adding and removing
species from ecological communities and hence have impor-
tant application for management actions such as enabling
lethal control, enhancing protection and conducting reintro-
ductions (Wallach et al. 2010; Ritchie et al. 2012; Ripple
et al. 2014; Doherty et al. 2015). For example, the structural
density of a network can predict the tendency of a given
ecosystem to colonization, population increases and declines,
and extinctions (Lurgi et al. 2014). Overall, the application
of network analysis is a powerful way to conceptualize nat-
ure not only by its species, but also by the architecture of its
interactions.
Acknowledgements
We thank the reviewers fo r helpful comments.
Data accessibility
This manuscript does not include any data.
References
Allesina, S. & Tang, S. (2012) Stability criteria for complex ecosystems. Nature,
483, 205208.
Atwood, T.B., Connolly, R.M., Ritchie, E.G., Lovelock, C.E., Heithaus, M.R.,
Hays, G.C., Fourqurean, J.W. & Macreadie, P.I. (2015) Predators help protect
carbon stocks in blue carbon ecosystems. Nature Climate Change,5,1038
1045.
Bascompte, J. (2009) Disentangling the web of life. Science,325, 416419.
Bascompte, J. & Stouffer, D.B. (2009) The assembly and disassembly of ecologi-
cal networks. Philosophical Transactions of the Royal Society of London. Series
B, Biological Sciences,364, 17811787.
Chapron, G., Kaczensky, P., Linnell, J.D.C., von Arx, M., Huber, D., Andr
en,
H. et al. (2014) Recovery of large carnivores in Europe’s modern human-
dominated landscapes. Science,346,15171519.
Colman,N.J.,Gordon,C.E.,Crowther,M.S.&Letnic,M.(2014)Lethal
control of an apex predator has unintended cascading effects on forest
mammal assemblages. Proceedings of the Royal Society B: Biological
Sciences,281, 20133094.
Creel, S. & Christianson, D. (2008) Relationships between direct predation and
risk effects. Trends in Ecology & Evolution,23, 194201.
Dekker, A. (2005) Conceptual distance in social network analysis. Journal of
Social Structure,6, 31.
Dickman, C.R., Glen, A.S., Jones, M.E., Soul
e,M.E.,Ritchie,E.G.&Wal-
lach, A.D. (2014) Strongly interactive carnivore species: maintaining and
restoring ecosystem function. Carnivores of Australia: Past, Present and
Future (eds A.S. Glen & C.R. Dickman), pp. 301322. CSIRO Publishing,
Melbourne, Vic., Australia.
Doherty, T.S., Dickman, C.R., Nimmo, D.G. & Ritchie, E.G. (2015) Multi-
ple threats, or multiplying the threats? Interactions between invasive
predators and other ecological disturbances. Biological Conservation,190,
6068.
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society, Methods in Ecology and Evolution
Trophic cascades in 3D 7
Estes, J.A., Terborgh, J., Brashares, J.S., Power, M.E., Berger, J., Bond, W.J.
et al. (2011) Trophic downgrading of planet Earth. Science,333,301306.
Fleming, P.A., Anderson, H., Prendergast, A.S., Bretz, M.R., Valentine, L.E. &
Hardy, G.E. (2014) Is the loss of Australian digging mammals contributing to
a deterioration in ecosystem functio n? Mammal Review,44,94108.
Glen, A.S. & Dickman, C.R. (2005) Complex interactions among mammalian
carnivores in Australia, and their implications for wildlife management.
Biological Reviews,80, 387.
Hairston, N.G., Smith, F.E. & Slobodkin, L.B. (1960) Community structure,
population control, and competition. American Naturalist,94, 421425.
Ings, T.C., Montoya, J.M., Bascompte, J., Bluthgen, N., Brown, L., Dormann,
C.F. et al. (2009) Ecological networks beyond food webs. Journal of Animal
Ecology,78,253269.
James, A.I., Eldridge, D.J. & Hill, B.M. (2009) Foraging animals create fertile
patches in an Australian desert shrubland. Ecography,32, 723732.
Johnson, C. (2006) Australia’s Mammal Extinctions: A 50000 Year History.Cam-
bridge University Press, Port Melbourne, Vic., Australia.
Johnson, C.N. & Isaac, J.L. (2009) Body mass and extinction risk in Aus-
tralian marsupials: the ‘critical weight range’ revisited. Austral Ecology,34,
3540.
K
efi, S., Berlow, E.L., Wieters, E.A., Navarrete, S.A., Petchey, O.L., Wood, S.A.
et al. (2012) More than a meal... integrating non-feeding interactions into
food webs. Ecology Letters,15,291300.
Knight, T.M., McCoy, M.W., Chase, J.M., McCoy, K.A. & Holt, R.D. (2005)
Trophic cascades across ecosystems. Nature,437,880883.
Knight, T.M., Chase, J.M., Hillebrand, H. & Holt, R.D. (2006) Predation on
mutualists can reduce the strength of trophic cascades. Ecology Letters,9,
11731178.
Letnic, M. & Crowther, M.S. (2013) Patterns in the abundance of kangaroo pop-
ulations in arid Australia are consistent with the exploitation ecosystems
hypothesis. Oikos,122, 761769.
Letnic, M., Ritchie, E.G. & Dickman, C.R. (2012) Top predators as biodiversity
regulators: the dingo Canis lupus dingo as a case study. Biological Reviews of
the Cambridge Philosophical Society,87, 390413.
Lurgi, M., Galiana, N., L
opez, B.C., Joppa, L. & Montoya, J.M. (2014) Network
complexity and species traits mediate the effects of biological invasions on
dynamic food webs. Population Dynamics,2,36.
McAlpine, C.A., Syktus, J., Ryan, J.G., Deo, R.C., McKeo n, G.M., McGowan,
H.A. & Phinn, S.R. (2009) A continent under stress: interactions, feedbacks
and risks associated with impact of modified land cover on Australia’s c limate.
Global Change Biology,15, 22062223.
Montoya, J.M., Pimm, S.L. & Sole, R.V. (2006) Ecological networks and their
fragility. Nature,442, 259264.
Newsome, T.M., Ballard, G.A., Crowther, M.S., Dellinger, J.A., Fleming, P.J.,
Glen, A.S. et al. (2015) Resolving the value of the dingo in ecological restora-
tion.Restoration Ecology,23, 201208.
Pascual, M. & Dunne, J.A. (2006) From small to large ecological networks in a
dynamic world. Ecological Networks: Linking Structure to Dynamics in Food
Webs (eds M. Pascual & J.A. Dunne),pp. 324. Oxford University Press, New
York, NY, USA.
Pimm, S.L., Lawton, J.H. & Cohen, J.E. (1991) Food web patterns and their con-
sequences. Nature,350,669674.
Potvin, F., Breton, L., Pilon, C. & Macquart, M. (1992) Impact of an experimen-
tal wolf reduction on beaver in Papineau-Labelle Reserve, Quebec. Canadian
Journal of Zoology,70, 180183.
Ripple, W.J., Estes, J.A., Beschta, R.L., Wilmers, C.C., Ritchie, E.G., Hebble-
white, M. et al. (2014) Status and ecological effects of the world’s largest carni-
vores. Science,343, 1241484.
Ritchie, E.G. & Johnson, C.N. (2009) Predator interactions, mesopredator
release and biodiversity conservation. Ecology Letters,12, 982998.
Ritchie, E.G., Elmhagen, B., Glen, A.S., Letnic, M., Ludwig, G. & McDonald ,
R.A. (2012) Ecosystem restoration with teeth: what role for predators? Trends
in Ecology & Evolution,27, 265271.
Rosell, F. & Sanda, J. (2006) Potential risks of olfactory signaling: the effect of
predators on scent marking by beavers. Behavioral Ecology,17, 897904.
S
aterberg, T., Sellman, S. & Ebenman, B. (2013) High frequency of functional
extinctions in ecological networks. Nature,499,468470.
Southgate, R., Paltridge, R., Masters, P. & Carthew, S. (2007) Bilby distribution
and fire: a test of alternative mod els of habitat suitability in the Tanami Desert,
Australia. Ecography,30, 759776.
Standish, R.J., Hobbs, R.J., Mayfield, M.M., Bestelmeyer, B.T., Suding, K.N.,
Battaglia, L.L. et al. (2014) Resilience in ecology: abstraction, distraction, or
where the action is? Biological Conservation,177,4351.
Thompson, R.M., Brose, U., Dunne, J.A., Hall, R.O. Jr, Hladyz, S., Kitching,
R.L. et al. (2012) Food web s: reconciling the structure and function of biodi-
versity. Trends in Ecology & Evolution,27, 689697.
Wallach, A.D., Johnson, C.N., Ritchie, E.G. & O’Neill, A.J. (2010) Predator con-
trol promotes invasive dominated ecological states. Ecology Letters,13, 1008
1018.
Wallach, A.D., Shanas, U., Izhaki, I., Toms, J.D. & Ripple, W.J. (2015) What is
an apex predator? Oikos,124, 14531461.
Wilmers, C.C. & Getz, W.M. (2005) Gray wolves as climate change buffers in
Yellowstone. PLoS Biology,3, e92.
Wolf, E.C., Cooper, D.J. & Hobbs, N.T. (2007) Hydrologic regime and herbivory
stabilize an alternative state in Yellowstone National Park. Ecological Applica-
tions,17, 15721587.
Received 14 June 2016; accepted 12 September 2016
Handling Editor: Diana Fisher
Supporting Information
Additional Supporting Information may be found online in the support-
ing information tab for this article:
Table S1. Method for assigning link weights from the results of mea-
sured species interactions.
Table S2. Description of studies used to assign link weight.
Table S3. Adjusted node and link weights entered into the network
model.
Video S1. File for Fig. 1c (online viewing only): Transition between the
network structures of two ecosystem states (ES) ES1 and ES2.
©2016 The Authors. Methods in Ecology and Evolution ©2016 British Ecological Society, Methods in Ecology and Evolution
8A. D. Wallach et al.
... According to focal study (Wallach et al., 2017), a set of rules were employed to translate the results from the selected studies on interaction strengths into link weights on a discrete scale (− 3 to +3) to represent strongly suppressive to strongly mutualistic interactions. Meantime, the link weights were determined by matching diets data for each species. ...
... When the weight of node X i is reduced, it is impacted in the network, and its link weight decreases as well (Tables2 and 3). Subsequently, this adjusted link weight determines the node weight of X j (Wallach et al., 2017). To be specific, a dominant or target species should be selected to give the node weight first (e.g., assigning the node weight of loach to 3, which feeding on zoobenthos). ...
Article
As amplified by climate changes and anthropogenic activities, mobile generalist species have been dominating the food web succession in a closed ecosystem. However, compared to marine ecosystem, the internal mechanism of generalist species affecting freshwater community were paid less attentions. In this study, the association of mobile generalist species and native apex predator was identified in Chenghai Lake, where was taken as the study case of a closed ecosystem to create the succession dynamic of food web from 1983 to 2019. Weighted food web analysis (WFWA) approach was applied to determine keystoneness and interaction strength of each species node, which is achieved through the quantification of the direct impacts on food web nodes from its biomass. As shown in the results, in the food web of 1983, there were nearly 25 nodes and 80 links, while in 2009, there were 22 nodes and 58 links, however, in 2019, there were 16 nodes and 25 links. In 2019, the three original specialized apex predator nodes were disappear. The keystoneness of each species was − 0.048, − 0.109 and − 0.742 respectively. During the nearly 40-year succession, including 13 native species had gone within 10 years, which were difficult to find its trace. The omnivorous alien species have occupied an absolute dominant niche, exhibiting the characteristics of mobile generalist species in 2019. The keystoneness of these species was − 0.194, − 0.235 and − 0.236 respectively. In general, from 1983 to 2019, the nodes have decreased by 36% and links have dropped by 68.75% on average in Chenghai Lake food web. In this study, (a)revealing the associations between mobile generalist species and native apex predators; (b) distinguishing the keystone species affecting food web structural; (c) developing food web analysis approach and determining the succession dynamic of Chenghai Lake. These findings improve the understanding of potential impacts from mobile generalist species in closed ecosystem.
... Ecological networks, and in particular food webs, provide a powerful unified framework to jointly address biodiversity and ecosystem functioning (Thompson et al., 2012), and guide the management of complex ecosystems (McDonald-Madden et al., 2016). Indeed, the food web concept encapsulates in a single framework cascading effects due to fluctuations in species' abundances, predator-prey dynamics, and emergent properties of direct and indirect trophic links at the scale of the whole network (Jordán et al., 2006;Wallach et al., 2017). The network structure is a major determinant of a food web capacity to withstand the spread of a perturbation (Albert et al., 2000;Solé and Montoya, 2001;Dunne et al., 2002b) and hence affects the robustness of ecosystems (Dunne et al., 2002b(Dunne et al., , 2004Tylianakis et al., 2007;Stouffer and Bascompte, 2011;Hattab et al., 2016;Robinson and Strauss, 2020). ...
Article
Full-text available
Management targets for biodiversity preservation are shifting from individual species to an ecosystem-wide focus. Indeed, the perturbation analysis of interaction networks, such as food webs, better captures the response of biodiversity to environmental pressures than single-species considerations. Here we propose a framework that examines food web robustness to a given perturbation based on life history traits and the topology of the food web, at different scales: local (species), intermediate (species directly linked together by a trophic interaction), and global (food web). Applying this framework to the Celtic Sea, a historically exploited fishing ground, we showed that the species sensitive to fishing were not the most central (i.e. with many interaction links, estimated based on eigenvector centrality) and that there is no both highly sensitive and exposed species to fishing. We then investigated how the loss of central, sensitive and exposed species to fishing could impact the robustness of the food web. We showed that the food web was the least robust to the simulated loss of species with many predators (i.e. forage species) and most exposed to fishing pressure, indicating that conservation priority could be focused on these species. Estimating species’ sensitivity to fishing was insufficient to predict food web robustness since the simulated removal of the most sensitive species led to a robustness level similar to that of a random removal sequence. Unlike what is often documented, the network appeared relatively robust to the simulated loss of the most central species, due notably to their implication in redundant trophic interactions and the fact that their disappearance increases modularity. This suggests that species-level metrics such as centrality should be completed by analysis at the scale of the whole food web to prioritize species conservation.
... Enfin, l'invasion de EEE de haut niveau trophique peut avoir des répercussions importantes dû à leur rôle structurant au sein des réseaux par prédation et compétition pouvant se répercuter sur d'autres niveaux trophiques par le biais de cascades « top-down » (Gozlan et al. 2010 ;Cucherousset et Olden 2011 ;Wallach et al. 2017 de Gallardo et al. 2015 et complétée ...
Thesis
Full-text available
Invasive species represent a growing threat to biodiversity and ecosystems worldwide and predicting their impacts on invaded ecosystems represents one of the major current challenges, particularly in the context of global change, promoting displacement of species out of their native range. The European catfish (Silurus glanis), the largest European freshwater fish, native from Asia and Eastern Europe, has extensively been proliferating across the continent since the 20th century. Indeed, its wide physiological, trophic and behavioral plasticity represents a major asset for establishing a sustainable presence in various ecosystems. S. glanis recently invaded large peri-Alpine lakes (Lake Geneva and Lake Bourget) raising concerns about its impacts on resident species in these ecosystems. This PhD thesis falls within this context and consisted in investigating (i) the trophic ecology of the European catfish in peri-Alpine lakes and (ii) its impacts on food webs and on resident species at both species and multispecies levels. This work is part of the SILAC research program founded by the Region Auvergne-Rhône-Alpes. The study of the European catfish diet in peri-Alpine lakes highlighted its high reliance on littoral prey (e.g., speeny-cheek crayfish Orconectes limosus and roach Rutilus rutilus) but also on prey from habitats inherent to peri-Alpine lakes such as deep benthic (e.g., burbot Lota lota) and pelagic prey (e.g., whitefish Coregonus lavaretus). An allometric niche model (aNM), supporting the reconstruction of trophic interactions in peri-Alpine lakes, was then elaborated and applied to a body size gradient representing the European catfish ontogeny to infer its trophic interactions both as predator and prey toward resident species. This step underlined its apex position in the food web reached as soon as individuals measured 60 cm. In addition, the consideration of another exotic species (Hemimysis anomala) indicated their potential additive impacts in the ecosystem. Trophic interactions inferred by the aNM were then coupled to a multispecies dynamic model to simulate the consequences of the European catfish (according to different body sizes) on the resident species abundances. The results indicated major negative impacts on the abundances of large invertebrates and meso-predators caused by S. glanis predation, but also a set of indirect positive and negative impacts on a large number of species characterizing top-down trophic cascades. Finally, the investigation of the vulnerability of the European whitefish (a Salmonidae with a high economic value) toward S. glanis allowed to identify a higher predation risk for the young-of-the year whitefish than for other whitefish life stages, due to their optimal body size for being predated by the European catfish as well as their high depth overlapping, particularly between the end of the spring and the end of the summer when the catfish energy demand is the highest. These results obtained on a period of recent colonization of the European catfish (< 10 years) constitute a robust referential for future investigations on the evolution of the population characteristics and its impacts in peri-Alpine lakes in longer term. The different methodologies developed during this PhD thesis contribute to the toolbox for a better assessment of invasive species impacts in aquatic ecosystems. Key words: invasive species, trophic ecology, food webs, impact assessment, population dynamics
... We state four hypotheses closely related to each other concerning predator-prey dynamics (Hypothesis 1), climate change (Hypothesis 2), intraguild competition (Hypothesis 3) and human impact (Hypothesis 4). Hypothesis 1 suggests that with changes in the carnivore community (e.g. the decrease in leopard population) we will observe changes in the frequency of prey (e.g. an increase in the frequency of leopards' main prey species, the Nubian ibex and the rock hyrax) (Wallach et al., 2017;Winnie and Creel, 2017). An alternative hypothesis (Hypothesis 2) to explain potential changes in large mammal herbivores is climate change: we expect that the increasing trend in heat and aridity in the region towards the late Holocene drier period would adversely affect large herbivore demography (Broughton et al., 2008); this may, in turn, influence predator density. ...
Article
Full-text available
Long temporal records of Holocene wild mammal communities are essential to examine the role of human impacts and climatic fluctuations in the configuration of modern ecosystems. We show that such records can be assembled through extensive radiocarbon dating of faunal remains obtained from biogenic cave deposits. We dated 110 mammalian remains from 19 different cave sites in the Judean Desert. We use the dates in combination with archaeological survey data and bone collagen/apatite δ¹³C values to study faunal succession in the context of Holocene climate change and human settlement history in the region. Our results suggest a change in the late Holocene, expressed in fewer observations of Arabian leopard (Panthera pardus nimr) and gazelle (Gazella spp.), and an increase of Syrian striped hyena (Hyaena hyaena syriaca), fox (Vulpes spp.), Nubian ibex (Capra nubiana) and rock hyrax (Procavia capensis); suids (Sus scrofa) appear for the first time. According to the data distribution, however, the probability of finding a bone diminishes exponentially with time, which implies that the Judean Desert cave paleozoological record is temporally biased. The weight of evidence ultimately favors an explanation of the observed patterns as the consequence of a combined anthropogenic and climatic impact on local food webs.
... For example, millions of springbok once traversed the semi-arid rangelands of South Africa, following the seasonal distribution of rainfall and forage (Prugh et al., 2009). The replacement of migrating wild herbivores with permanently resident domestic herbivores has altered patterns of nutrient cycling (Doughty et al., 2016;Lundgren et al., 2020;Western, 1975;Westman, 1990), soil turnover and compaction, and trophic cascades (Wallach et al., 2017a(Wallach et al., , 2009. Uncoupling the ecological processes that made these landscapes resilient to extended dry periods has destabilised these rangeland systems Fensham, 2019, 2013), making them more susceptible to aridification under rapid climate change (Walker et al., 2018). ...
Article
While rangeland ecosystems are globally important for livestock production, they also support diverse wildlife assemblages and are crucial for biodiversity conservation. As rangelands around the world have become increasingly degraded and fragmented, rethinking farming practice in these landscapes is vital for achieving conservation goals, rangeland recovery, and food security. An example is reinstating livestock shepherding, which aims to recouple grazing regimes to vegetation conditioned to semi-arid climates and improve productivity by reducing overgrazing and rewiring past ecological functions. Tracking the large-scale ecosystem responses to shifts in land management in such sparsely vegetated environments have so far proven elusive. Therefore, our goal was to develop a remote tracking method capable of detecting vegetation changes and environmental responses on rangeland farms engaging in contrasting farming practices in South Africa: wildlife friendly farming (WFF) implementing livestock shepherding with wildlife protection, or rotational grazing livestock farming with wildlife removal. To do so, we ground-truthed Sentinel-2 satellite imagery using drone imagery and machine learning methods to trace historical vegetation change on four farms over a four-year period. First, we successfully classified land cover maps cover using drone footage and modelled vegetation cover using satellite vegetation indices, achieving 93.4% accuracy (к = 0.901) and an r-squared of 0.862 (RMSE = 0.058) respectively. We then used this model to compare the WFF farm to three neighbouring rotational grazing farms, finding that satellite-derived vegetation productivity was greater and responded more strongly to rainfall events on the WFF farm. Furthermore, vegetation cover and grass cover, patch size, and aggregation were greater on the WFF farm when classified using drone data. Overall, we found that remotely assessing regional environmental benefits from contrasting farming practices in rangeland ecosystems could aid further adoption of wildlife-friendly practices and help to assess the generality of this case study.
... Ecological network enables to evaluate the vulnerability of ecosystems to a perturbation through the study of the changes in the structure of the network (Hattab et al., 2016;Robinson & Strauss, 2020;Stouffer & Bascompte, 2011;Tylianakis, Tscharntke, & Lewis, 2007). Its use to assess ecosystem state has increased in the recent years as it allows considering in a single framework the effects of fluctuation in species' abundance and their preys and predators, but also on indirectly linked species and the whole network itself (Jordán et al., 2006;Wallach et al., 2017). ...
... Ecological network enables to evaluate the vulnerability of ecosystems to a perturbation through the study of the changes in the structure of the network (Hattab et al., 2016;Robinson & Strauss, 2020;Stouffer & Bascompte, 2011;Tylianakis, Tscharntke, & Lewis, 2007). Its use to assess ecosystem state has increased in the recent years as it allows considering in a single framework the effects of fluctuation in species' abundance and their preys and predators, but also on indirectly linked species and the whole network itself (Jordán et al., 2006;Wallach et al., 2017). ...
Thesis
Full-text available
Marine ecosystems face many pressures, the main ones being fishing and climate change. These pressures do not have the same consequences on the different facets of biodiversity, e.g. on taxa, traits and networks, and their study is necessary for the implementation of an ecosystem approach to fisheries. The Celtic Sea is a very important fishing area for European fisheries, with a specific richness greater than adjacent areas. Based on a time series from 2000 to 2016, I have studied the spatio-temporal dynamics of 101 species of the ecosystem. Over this period, taxonomic biodiversity was stable, and the importance of fishing as a structuring variable for communities decreased after 2009, and was lower than the environment (depth, chlorophyll a concentration and temperature). Following a local decrease in fishing pressure, the biomass of species with life history strategies characterised by a low renewal rate and present in the central Celtic Sea, increased. In particular, the increase in the biomass of six species of elasmobranchs has been linked to the creation of a seasonal fishing closure in the area. This vision of fishing sensitivity was enriched with data from the exploitation of underwater videos. This made it possible to highlight the presence of sensitive areas in the western Celtic Sea, along the continental slope. In addition, the study of interaction networks linking species, and in particular trophic networks, allowed considering the dynamics of the entire ecosystem based on its components. In the Celtic Sea, the sensitivity of species to fishing does not imply the sensitivity of the network to this pressure. On the other hand, the trophic network is not very robust to the loss of the 7% of the most exploited species and basal species.
... The complex nature of invasive species management suggests that trophic cascades are a likely outcome, particularly where mesopredators are present (Caut et al. 2007, Rayner et al. 2007, Raymond et al. 2011, Wallach et al. 2017. The eradication or control of feral cats has attracted considerable focus globally with their impacts on threatened fauna becoming increasingly well understood (Doherty et al. 2017). ...
Article
Managing ecosystems in the face of complex species interactions, and the associated uncertainty, presents a considerable ecological challenge. Altering those interactions via actions such as invasive species management or conservation translocations can result in unintended consequences, supporting the need to be able to make more informed decisions in the face of this uncertainty. We demonstrate the utility of ecosystem models to reduce uncertainty and inform future ecosystem management. We use Phillip Island, Australia, as a case study to investigate the impacts of two invasive species management options and consider whether a critically endangered mammal is likely to establish a population in the presence of invasive species. Qualitative models are used to determine the effects of apex predator removal (feral cats) and invasive prey removal (rabbits, rats, and mice). We extend this approach using Ensemble Ecosystem Models to consider how suppression, rather than eradication influences the species community; and consider whether an introduction of the critically endangered eastern barred bandicoot is likely to be successful in the presence of invasive species. Our analysis revealed the potential for unintended outcomes associated with feral cat control operations, with rats and rabbits expected to increase in abundance. A strategy based on managing prey species appeared to have the most ecosystem‐wide benefits, with rodent control showing more favourable responses than a rabbit control strategy. Eastern barred bandicoots were predicted to persist under all feral cat control levels (including no‐control). Managing ecosystems is a complex and imprecise process. However, Qualitative Modelling and Ensemble Ecosystem Modelling address uncertainty and are capable of improving and optimising management practices. Our analysis shows that the best conservation outcomes may not always be associated with the top –down control of apex predators, and land managers should think more broadly in relation to managing bottom‐up processes as well. Challenges faced in continuing to conserve biodiversity mean new, bolder, conservation actions are needed. We suggest that endangered species are capable of surviving in the presence of feral cats, potentially opening the door for more conservation translocations.
... Ecological network enables to evaluate the vulnerability of ecosystems to a perturbation through the study of the changes in the structure of the network (Tylianakiset al. 2007;Stouffer & Bascompte 2011;Hattab et al.2016;Robinson & Strauss 2020). Its use to assess ecosystem state has increased in the recent years as it allows considering in a single framework the effects of fluctuation in species' abundance and their preys and predators, but also on indirectly linked species and the whole network itself (Jordán et al. 2006;Wallach et al. 2017). ...
Preprint
The analysis of the dynamics of interaction networks (i.e. trophic webs) better capture the state of ecosystem facing a perturbation than individual species dynamics could. We propose a framework that examines network robustness to a given perturbation at the local (species), mesoscale (species directly linked together) and global (network) level, based on traits and the topology of the network. Using the Celtic Sea as an example, we showed that the network was the least robust to the simulated loss of forage taxa and the most exposed taxa to fishing pressure, indicating conservation priority could be focused on these taxa. However estimating the sensitivity to fishing at the taxa ‘level was insufficient to predict the robustness of the network. The network appeared relatively robust to the simulated loss of the most central taxa, suggesting that mesoscale metrics such as centrality, although widely used, are not always adapted to prioritize species conservation.
Preprint
Full-text available
Climate change alters ecological communities through effects on individual species and interactions between species. The impacts of climate change may be buffered by community diversity: diverse communities may be more resistant to climate-driven perturbations than simple communities. Here, we assess how diversity influences long-term thermal niche variation in communities under climate change. We use 50-year continental-scale data on bird communities during breeding and non-breeding seasons to quantify the communities’ thermal variability. Thermal variability is measured as the temporal change in community’s average temperature niche that indicates its response to climate change. Then, we study how the thermal variability varies as a function of taxonomic, functional, and evolutionary diversity using linear models. We find that communities with low thermal niche variation have higher functional diversity, with this pattern being measurable in the non-breeding but not in the breeding season. Given the increasing seasonal variation in the future climate change effects, the differences in bird communities’ thermal variability between breeding and non-breeding seasons may grow wider. Importantly, our results suggest that functionally diverse wildlife communities can mitigate negative effects of climate change by means of hindering changes in thermal niche variability, underscoring the importance of addressing the climate and biodiversity crises together.
Article
Full-text available
Understanding the mechanisms by which climate and predation patterns by top predators co-vary to affect community structure accrues added importance as humans exert growing influence over both climate and regional predator assemblages. In Yellowstone National Park, winter conditions and reintroduced gray wolves (Canis lupus) together determine the availability of winter carrion on which numerous scavenger species depend for survival and reproduction. As climate changes in Yellowstone, therefore, scavenger species may experience a dramatic reshuffling of food resources. As such, we analyzed 55 y of weather data from Yellowstone in order to determine trends in winter conditions. We found that winters are getting shorter, as measured by the number of days with snow on the ground, due to decreased snowfall and increased number of days with temperatures above freezing. To investigate synergistic effects of human and climatic alterations of species interactions, we used an empirically derived model to show that in the absence of wolves, early snow thaw leads to a substantial reduction in late-winter carrion, causing potential food bottlenecks for scavengers. In addition, by narrowing the window of time over which carrion is available and thereby creating a resource pulse, climate change likely favors scavengers that can quickly track food sources over great distances. Wolves, however, largely mitigate late-winter reduction in carrion due to earlier snow thaws. By buffering the effects of climate change on carrion availability, wolves allow scavengers to adapt to a changing environment over a longer time scale more commensurate with natural processes. This study illustrates the importance of restoring and maintaining intact food chains in the face of large-scale environmental perturbations such as climate change.
Article
Full-text available
Biological invasions are a major threat to natural communities worldwide. While several species traits have been identified as important determinants of invasion success, a systematic exploration of the effects of invasions on native communities, and the role of species and community features on community robustness in the face of invasion is lacking. We present a theoretical approximation considering food web structure and species population dynamics to study the effects of invasions on complex food webs. We find that less complex (i.e. less connected) food webs are more resistant to invasions. Simulated invasions promote profound changes in several food web properties and stability measures, such as decreases in modularity and the number of food chains from basal to top species; and a decoupling of community- and population-level temporal variability. Additionally, species traits such as body size and diet breadth are strong determinants of invasion success across several trophic levels, with larger and more generalist species being more successful invaders in general. Our work complements species-centred invasion studies by adding a more holistic and systematic perspective to the study of invasions on species interaction networks.
Article
Full-text available
There is global interest in restoring populations of apex predators, both to conserve them and to harness their ecological services. In Australia, reintroduction of dingoes (Canis dingo) has been proposed to help restore degraded rangelands. This proposal is based on theories and the results of studies suggesting that dingoes can suppress populations of prey (especially medium- and large-sized herbivores) and invasive predators such as red foxes (Vulpes vulpes) and feral cats (Felis catus) that prey on threatened native species. However, the idea of dingo reintroduction has met opposition, especially from scientists who query the dingo's positive effects for some species or in some environments. Here, we ask 'what is a feasible experimental design for assessing the role of dingoes in ecological restoration?' We outline and propose a dingo reintroduction experiment-one that draws upon the existing dingo-proof fence-and identify an area suitable for this (Sturt National Park, western New South Wales). Although challenging, this initiative would test whether dingoes can help restore Australia's rangeland biodiversity, and potentially provide proof-of-concept for apex predator reintroductions globally.
Article
Full-text available
Large ‘apex’ predators influence ecosystems in profound ways, by limiting the density of their prey and controlling smaller ‘mesopredators’. The loss of apex predators from much of their range has lead to a global outbreak of mesopredators, a process known as ‘mesopredator release’ that increases predation pressure and diminishes biodiversity. While the classifications apex- and meso-predator are fundamental to current ecological thinking, their definition has remained ambiguous. Trophic cascades theory has shown the importance of predation as a limit to population size for a variety of taxa (top–down control). The largest of predators however are unlikely to be limited in this fashion, and their densities are commonly assumed to be determined by the availability of their prey (bottom–up control). However, bottom–up regulation of apex predators is contradicted by many studies, particularly of non-hunted populations. We offer an alternative view that apex predators are distinguishable by a capacity to limit their own population densities (self-regulation). We tested this idea using a set of life-history traits that could contribute to self-regulation in the Carnivora, and found that an upper limit body mass of 34 kg (corresponding with an average mass of 13–16 kg) marks a transition between extrinsically- and self-regulated carnivores. Small carnivores share fast reproductive rates and development and higher densities. Large carnivores share slow reproductive rates and development, extended parental care, sparsely populated territories, and a propensity towards infanticide, reproductive suppression, alloparental care and cooperative hunting. We discuss how the expression of traits that contribute to self-regulation (e.g. reproductive suppression) depends on social stability, and highlight the importance of studying predator–prey dynamics in the absence of predator persecution. Self-regulation in large carnivores may ensure that the largest and the fiercest do not overexploit their resources.
Article
Full-text available
There is global interest in restoring populations of apex predators, both to conserve them and to harness their ecological services. In Australia, reintroduction of dingoes (Canis dingo) has been proposed to help restore degraded rangelands. This proposal is based on theories and the results of studies suggesting that dingoes can suppress populations of prey (especially medium- and large-sized herbivores) and invasive predators such as red foxes (Vulpes vulpes) and feral cats (Felis catus) that prey on threatened native species. However, the idea of dingo reintroduction has met opposition, especially from scientists who query the dingo’s positive effects for some species or in some environments. Here, we ask ‘what is a feasible experimental design for assessing the role of dingoes in ecological restoration?’We outline and propose a dingo reintroduction experiment—one that draws upon the existing dingo-proof fence—and identify an area suitable for this (Sturt National Park, western New South Wales). Although challenging, this initiative would test whether dingoes can help restore Australia’s rangeland biodiversity, and potentially provide proof-of-concept for apex predator reintroductions globally.
Article
Predators continue to be harvested unsustainably throughout most of the Earth’s ecosystems. Recent research demonstrates that the functional loss of predators could have far-reaching consequences on carbon cycling and, by implication, our ability to ameliorate climate change impacts. Yet the influence of predators on carbon accumulation and preservation in vegetated coastal habitats (that is, salt marshes, seagrass meadows and mangroves) is poorly understood, despite these being some of the Earth’s most vulnerable and carbon-rich ecosystems. Here we discuss potential pathways by which trophic downgrading affects carbon capture, accumulation and preservation in vegetated coastal habitats. We identify an urgent need for further research on the influence of predators on carbon cycling in vegetated coastal habitats, and ultimately the role that these systems play in climate change mitigation. There is, however, sufficient evidence to suggest that intact predator populations are critical to maintaining or growing reserves of ‘blue carbon’ (carbon stored in coastal or marine ecosystems), and policy and management need to be improved to reflect these realities.
Article
Invasive species have reshaped the composition of biomes across the globe, and considerable cost is now associated with minimising their ecological, social and economic impacts. Mammalian predators are among the most damaging invaders, having caused numerous species extinctions. Here, we review evidence of interactions between invasive predators and six key threats that together have strong potential to influence both the impacts of the predators, and their management. We show that impacts of invasive predators can be classified as either functional or numerical, and that they interact with other threats through both habitat- and community-mediated pathways. Ecosystem context and invasive predator identity are central in shaping variability in these relationships and their outcomes. Greater recognition of the ecological complexities between major processes that threaten biodiversity, including changing spatial and temporal relationships among species, is required to both advance ecological theory and improve conservation actions and outcomes. We discuss how novel approaches to conservation management can be used to address interactions between threatening processes and ameliorate invasive predator impacts.