Ecology Letters, (2013) 16: 664–669
LETTER
Dirk Sanders,1,2* Louis Sutter1 and
F. J. Frank van Veen2
doi: 10.1111/ele.12096
The loss of indirect interactions leads to cascading extinctions
of carnivores
Abstract
Species extinctions are biased towards higher trophic levels, and primary extinctions are often followed by
unexpected secondary extinctions. Currently, predictions on the vulnerability of ecological communities to
extinction cascades are based on models that focus on bottom-up effects, which cannot capture the effects
of extinctions at higher trophic levels. We show, in experimental insect communities, that harvesting of single carnivorous parasitoid species led to a significant increase in extinction rate of other parasitoid species,
separated by four trophic links. Harvesting resulted in the release of prey from top-down control, leading
to increased interspecific competition at the herbivore trophic level. This resulted in increased extinction
rates of non-harvested parasitoid species when their host had become rare relative to other herbivores. The
results demonstrate a mechanism for horizontal extinction cascades, and illustrate that altering the relationship between a predator and its prey can cause wide-ranging ripple effects through ecosystems, including
unexpected extinctions.
Keywords
Aphids, community stability, indirect effects, non-trophic interactions, parasitoids, resource competition,
secondary extinctions, species loss.
Ecology Letters (2013) 16: 664–669
It is argued that we are entering Earth’s sixth mass extinction
(Barnosky et al. 2011). The biodiversity loss as a result of human
activities can lead to cascades of secondary extinctions (Paine 1966;
Borrvall & Ebenman 2006; Montoya et al. 2006). Predicting secondary extinctions is difficult and requires an understanding of the
mechanisms by which the impact of species loss is transmitted
through the network of species (Ives & Cardinale 2004). Indirect
interactions, for example, via chains of feeding interactions, can play
a dominant role in structuring communities (Wootton 1994; Bukovinszky et al. 2008) and therefore in propagating the effects of
human-induced species loss. Janzen (1974) raised the spectre of the
‘most insidious sort of extinction, the extinction of ecological interactions’ and indeed global warming may lead to the large-scale
extinction of plant pollinator interactions (Memmott et al. 2007) and
overfishing can have a wide-ranging impact on food web dynamics
(Cushing 1988).
Top-down population regulation by carnivores can promote the
coexistence of multiple species at the prey trophic level by reducing
interspecific competition (Paine 1966). Where carnivore species specialise on different, but competing prey, this can lead to indirect
positive interactions between consumers (Sanders & van Veen
2012) and it has been suggested that this could be an important factor in maintaining biodiversity at higher trophic levels (Dodson
1970; Abrams & Nakajima 2007). Species in higher trophic positions experience a greater extinction frequency, which is correlated
with their larger body sizes, lower abundances and a result of cumulative effects of disturbance to species lower down the food chain
(McKinney 1997; Petchey et al. 1999; Purvis et al. 2000). Therefore,
top predators are particularly vulnerable to human-induced disturbances on ecosystems (Borrvall & Ebenman 2006), such as harvesting (Jackson et al. 2001; Myers & Worm 2003) and climate change
(Voigt et al. 2003). These perturbations may therefore have especially strong effects on top-consumer extinctions, effects that can
ripple through entire food webs, multiplying extinction risks along
the way (Zarnetske et al. 2012). Removing top-consumers leads to
disproportionate changes in community composition across trophic
levels (Schmitz et al. 2003; Myers et al. 2007; Heithaus et al. 2008)
by cascading down the food web (vertical interactions). An example
are widespread reductions of kelp forests in the Northern Hemisphere following the removal of top predators by fishing which was
coupled with population explosions of herbivores (Jackson et al.
2001). However, these kind of effects can also be transmitted via
indirect interactions between species at the same trophic level
(Sanders & van Veen 2012), proposing the risk of horizontal extinction cascades among top-consumers.
To test whether altering predator–prey interactions through harvesting of species at higher trophic levels leads to the loss of important indirect positive interactions and the extinction of other
carnivores, we assembled replicate communities in the laboratory
consisting of three herbivorous aphid species on a single shared
resource and three parasitoid wasp species, each specialising on a
different aphid species. We hypothesised that systematic harvesting
of one parasitoid would have a negative impact on the population
density and persistence of the remaining parasitoids as a result of
increased resource competition between their respective host species. We demonstrate that harvesting of one carnivore (parasitoid)
species leads to the extinction of others and that these effects are
transmitted via at least four trophic links. Furthermore, we show
1
2
INTRODUCTION
Community Ecology, Institute of Ecology and Evolution, Baltzerstrasse 6,
3012, Bern, Switzerland
Centre for Ecology & Conservation, College of Life and Environmental Sciences,
University of Exeter, Cornwall Campus, Penryn, Cornwall, TR10 9EZ, UK
*Correspondence: E-mail: [email protected]
© 2013 Blackwell Publishing Ltd/CNRS
Letter
Harvesting of carnivores leads to extinction cascades 665
that the extinctions are not just the result of chains of extinctions
along trophic links, but are also likely in part caused by changes in
the strength of non-trophic interference interactions. These results
illustrate that the effects of altering the relationship between a predator and its prey will not only cascade down the food chain but can
cause wide-ranging ripple effects through ecological communities,
including cascading extinctions within the top trophic level.
Experiment
We used plant–aphid–parasitoid communities which consisted of
bean plants (Vicia faba, L., var. the Sutton) as food resource for the
three aphid species Aphis fabae (Scopoli), Acyrthosiphon pisum (Harris)
and Megoura viciae (Buckton), each attacked by a specialist parasitoid,
(b)
14
16
18
20
22
(d)
8
10
1
M. viciae
12
14
16
18
20
22
14
16
18
20
22
12
14
16
18
20
22
12
14
16
18
20
22
A. megourae
8
(f)
12
10
2
A.pisum
8
10
1
5
20
5
100
20
1000
(e)
10
100
5
20
100
1000
Parasitoid abundance
(c)
Aphid abundance
8
100
12
20
10
L. fabarum
5
8
1
A. fabae
2
5
20
5
20
100
100
1000
(a)
MATERIAL AND METHODS
12
14
16
18
20
22
A.ervi
8
10
Week
No harvesting control
L. fabarum harvested
A. megourae harvested
A. ervi harvested
Week
L. fabarum A. megourae
A. fabae
M. viciae
A. ervi
A. pisum
V. faba
Figure 1 Populations dynamics in the experimental insect communities. (a) A. fabae, (b) L. fabarum, (c) M. viciae, (d) A. megourae, (e) A. pisum and (f) A.ervi. Presented are
means for species abundance (from seven replicates) and SE on a log scale.
© 2013 Blackwell Publishing Ltd/CNRS
666 D. Sanders, L. Sutter and F. J. Frank van Veen
which were Lysiphlebus fabarum (Marshall), Aphidius ervi (Haliday)
and Aphidius megourae (Stary) respectively (see Fig. 1). Prior to the
experiments, parasitoids and their hosts were maintained on bean
plants and kept in a controlled environment room at 20 °C, with a
16 : 8-h light:dark cycle.
The experimental units were 48 9 48 9 48 cm gauze population
cages containing eight pots with broad beans. The experiments were
initiated by the introduction of five wingless adults of each aphid
species into the experimental cages. To ensure overlapping parasitoid generations, 1 week and 2 weeks later, two mated parasitoid
females of each species were released into each cage. After 8 weeks
(four generations), we established four different treatments (a) no
harvesting control; (b) harvesting of L. fabarum; (c) harvesting of
A. megourae and (d) harvesting of A. ervi. Harvesting was done by
removing mummies of single parasitoid species with a daily effort
of ~15 min per cage with control cages being disturbed in a similar
way (search activity and inspecting each single pot with plants).
Each treatment was replicated seven times and treatments were
arranged in blocks within the controlled temperature room. Bean
plants were 2 weeks old when being placed inside the cages. The
plants were renewed by twice a week replacing the two oldest pots
with fresh ones containing two-week-old seedlings, taking care to
return all insects to the cage. Previous experiments have shown that
this allows for long term maintenance of populations but does not
prevent competition among the aphids (van Veen et al. 2005; Sanders & van Veen 2012). Once a week, during the 22 weeks of the
experiment, aphids and aphid mummies of each species were
counted on half of the plants (one plant from each age stage) in
each cage.
To understand the directions and strengths of competitive interactions between the three aphid species in the absence of parasitoids, we measured their population dynamics in a similar
experiment with the following treatments: each species on its own,
two species combinations and all three species together (total seven
different treatments with three replicates of each). This experiment
ran for 7 weeks until aphid populations had peaked after reaching
the carrying capacity.
Statistical analysis
To compare extinction probability between the different treatments
(non-harvesting control and harvesting of each of the parasitoid species), the number of weeks a parasitoid species persisted in each replicate was noted. Species that persisted until the end of the experiment
were treated as censored data. The Kaplan–Meier survival curves of
the three treatments were compared using a log-rank test as implemented in the R-function survdiff (Harrington & Fleming 1982).
First, we tested for a significant difference between all survival curves
for each species. In case this overall test was significant, we compared
single treatment-non-harvesting control pairs for the impact of harvesting single species on the persistence of other parasitoids.
The responses of aphid and parasitoid population dynamics to
the experimental treatments were analysed using linear mixed-effects
models with log transformed density data as dependent variable
assuming normal error distribution. Treatment was included as a
fixed factor. To account for systematic trends over time we
included week, and week squared as covariables. Replicate number
nested in block was used as random factors in the model. These
mixed effects techniques allow replicate time series to be analysed
© 2013 Blackwell Publishing Ltd/CNRS
Letter
taking into account the potential non-independence of repeated
measures. Because the residuals of this model showed a significant
partial temporal autocorrelation of order one in 14 of 28 replicates,
we included a first-order autoregression for the residuals in the
model. This was done using the function lme from the package
nlme by Pinheiro et al. (2011).
Host relative abundance was analysed using generalised linear
mixed models assuming a binomial error distribution and using the logit link function. The dependent variable was the bivariate variable
containing ‘abundance of aphid species i’ and ‘total aphid abundance abundance of aphid species i’, where ‘i’ can be the abundance of
A. fabae, M. viciae, or A. pisum. Treatment was included as fixed factor,
and week and week squared as covariates. The random factors were
block and replicate nested in block. We also included a random slope
for the week effect per replicate since residual analyses showed that
the linear trend over time differed between replicates. The inclusion
of replicate-specific slopes is also supported by the BIC (Bayesian
Information Criterion) (DBIC = 9.1). We further included an
observation level random factor to account for over-dispersion that
seemed to be substantial in our data (additional between-observation
variance was 2.06, compared to between-replicate variance of 0.24
and between-block variance of 0.007). Temporal autocorrelation was
negligible in this model (all partial autocorrelations below 0.2), and
therefore not included in the model. We used the function glmer from
the R-package lme4 (Bates et al. 2011) to fit this model. To obtain
95% credible intervals for the model predictions, we used Bayesian
methods as recommended by e.g. Bolker et al. (2009). To do so, we
used the function sim from the R-package arm (Gelman et al. 2012)
to draw a random sample of 1 000 values from the joint posterior distribution of the model parameters. From these 1 000 sets of model
parameters, 1 000 predicted values were calculated and their 2.5%
and 97.5% quantiles used as lower and upper limits of the 95% credible intervals.
The response of aphid species to the presence of other aphid
species in the competition experiment was analysed in the same way
as the aphid and parasitoid population dynamics in the main experiment. For each aphid species we only included the data from cages
in which the species was present (single species treatment, with each
of the other two species, and the combination of all three species).
All statistical analyses were performed in the open source software
R 2.13.1 (2011).
RESULTS
Harvesting experiment
Harvesting of parasitoids led to a strong decline in the density of
each target species in the different treatments when compared to
the full community: density of A. megourae was reduced by 81%
(t1,18 = 4.62 P = 0.0002), A. ervi by 79% (t1,18 = 6.09,
P < 0.0001), and L. fabarum by 76% (t1,18 = 4.31, P = 0.0004)
(Fig. 1b,d,f).
Harvesting of single parasitoid species led to extinctions of other
parasitoids that were separated by four trophic links. The overall
test for differences in survival curves was significant for A. megourae
(v2 3 = 6.9, P = 0.0264) and A. ervi (v2 3 = 8.2, P = 0.0194) but
not for L. fabarum (v2 3 = 4.8, P = 0.183). Population persistence
declined markedly for the parasitoid A. megourae when being harvested (v2 1 = 6.5, P = 0.0110) and in the L. fabarum-harvesting
Letter
Harvesting of carnivores leads to extinction cascades 667
(a)
ns
0.8
ns
0.0
0.4
ns
L. fabarum
10
14
18
22
0.8
(b)
0.4
ns
0.0
Proportion of replicates persisting
8
harvested (v2 1 = 7.3, P = 0.0067) and when one of the other parasitoid species, L. fabarum (v2 1 = 7.3, P = 0.0070) or A. megourae (v2
1 = 7.4, P = 0.0066), had been harvested, with no extinction events
in the non-harvesting treatment (Fig. 2). Thus, there was a clear
indirect effect of harvesting L. fabarum on A. ervi and A. megourae,
and of harvesting A. megourae on A. ervi.
The removal of parasitoid species initiated an increase in density
of their specific hosts (Fig. 1), which was 2.8-fold for A. pisum
(t1,18 = 2.65, P = 0.0161), 5-fold for M. viciae (t1,18 = 3.13,
P = 0.0058), and 3.1-fold for A. fabae (t1,18 = 2.40, P = 0.0272).
Total aphid densities almost doubled in harvesting treatments as
compared to the non-harvesting control. Hosts that were released
from top-down control dominated aphid communities relative to
the other species which was most pronounced for A. fabae
(z = 3.87, P < 0.001) and M. viciae (z = 4.81, P < 0.001) but not
significant, however with the same trend, for A. pisum (z = 1.38,
P = 0.1664) (see Appendix S1 in Supporting Information). Parasitoid extinctions occurred when their host had become rare relative
to the other aphid species. All extinction events for parasitoids were
recorded when their respective hosts had reached a 3.6 3.5%
mean relative abundance ( SE). Host species never went extinct.
*
*
A. megourae
8
10
14
18
22
(c)
0.8
**
**
0.0
0.4
**
Competition experiment
A. fabae was the competitively dominant species (no impact of
A. pisum t1,6 = 0.51 P = 0.6284, and M. viciae t1,6 = 0.08
P = 0.9395, and the combination of both species t1,6 = 1.98
P = 0.0954, see Appendix S2 A in Supporting Information), followed by M. viciae (A. fabae effect t1,6 = 3.20 P = 0.0185, no effect
of A. pisum t1,6 = 0.48 P = 0.6476, and the combination of both
species t1,6 = 0.05 P = 0.9595, see Appendix S2 B). A. pisum suffered in the presence of both other species (A. fabae effect
t1,6 = 2.48 P = 0.0479, M. viciae effect t1,6 = 2.62 P = 0.0395,
and in combination t1,6 = 3.19 P = 0.0188, see Appendix S2 C).
DISCUSSION
A.ervi
8
10
14
18
22
Week
No harvesting control
L. fabarum harvested
A. megourae harvested
A. ervi harvested
Figure 2 Persistence of experimental parasitoid populations of (a) L. fabarum, (b)
A. megourae and (c) A. ervi in full communities (black line) and after harvesting of
single parasitoid species (coloured lines). To allow all species to get established,
harvesting started at week 8 of the experiment (after four parasitoid generations).
treatment (v2 1 = 5.6, P = 0.0170) with two extinction events in the
non-harvesting treatment (Fig. 2). Harvesting A. ervi had no significant effect (v2 1 = 2, P = 0.158) on the persistence of A. megourae
although with a similar trend to the impact of harvesting L. fabarum
(Fig. 2). The persistence of A. ervi declined significantly when
In our experiment, vertical and horizontal species interactions transmitted the impact of harvesting through the whole system leading
to the extinction of other, indirectly linked, species at the top trophic level. Sustained harvesting of single parasitoid species markedly
reduced the persistence of other parasitoid species: harvesting
L. fabarum led to increased extinction rate of A. megourae and A. ervi;
harvesting A. megourae led to increased extinction rate of A. ervi.
Remarkably, these indirect effects on non-target species were of
similar strength to the effect of directly harvesting the affected species (Fig. 2b and c). This illustrates that altering the relationship
between a predator and its prey can cause unexpected extinctions
of indirectly linked species, including ones at the same trophic level
as the harvested species and that these effects can be very strong.
We expected the competitive exclusion of hosts, by competitors
released from top-down control (Sanders & van Veen 2012), to be
the cause of parasitoid extinction. However, only A. pisum showed
clear declines in population density following harvesting of its competitors’ parasitoids (Fig. 1e) and in no cases did any aphid population go extinct. So parasitoids went extinct while their hosts still
persisted. The extinctions of non-target parasitoids were most likely
mediated by changes in relative host densities, which affected parasitoid searching efficiency: It has repeatedly been observed that high
© 2013 Blackwell Publishing Ltd/CNRS
668 D. Sanders, L. Sutter and F. J. Frank van Veen
non-host relative density leads to reduced per capita attack rates for
parasitoids (Vos et al. 2001; van Veen et al. 2005; Meisner et al.
2007), probably due to masking of chemical cues, increased complexity of search space and generalised anti-predator behaviour of
non-hosts. Although this can have a stabilising effect on inherently
unstable host-parasitoid dynamics, when the effect is too strong it
can lead to parasitoid extinctions (Vos et al. 2001).
These kind of effects are not limited to host-parasitoid systems
and reduced attack or ingestion rates with higher diversity at the
resource trophic level, and lower resource relative density, have been
observed in a wide variety of systems, from marine predator–prey to
terrestrial plant–herbivore interactions (reviewed in van Veen &
Godfray 2012). A related mechanism in host-parasitoid systems may
operate when parasitoids cannot discriminate between host and nonhost species. In such a case, the non-host can act as a sink of parasitoid eggs, leading to a decrease in parasitoid efficiency with increasing
non-host density (Heimpel et al. 2003). We have never observed any
of the parasitoid species in our experimental system attempting to
oviposit in non-hosts so this latter mechanism cannot be responsible
for the observed extinctions. However, it may operate in other hostparasitoid systems, especially where novel unsuitable hosts are introduced into communities with na€ıve parasitoid populations.
Secondary extinctions themselves may lead to further extinctions.
It appears that the extinction of A. ervi following harvesting of
L. fabarum was the result of the initial decline of A. megourae and
therefore represents such an extinction cascade: in a previous
experiment, in which A. megourae and its host were not present, we
found no evidence for an indirect effect of L. fabarum on A. ervi
(Sanders & van Veen 2012). We suggest therefore that the increase in
A. fabae (Fig. 1a blue line), following the harvesting of its parasitoid
L. fabarum, was responsible for the rapid increase in extinction rate in
A. megourae (Fig. 2b) which in turn led to an increase in M. viciae
population densities (Fig. 1c, blue line) and the subsequent extinction
of A. ervi. A strong negative effect of M. viciae on A. ervi’s attack rate
on A. pisum has been demonstrated previously (van Veen et al. 2005).
On the basis of the above mentioned previous experiment with only
two species pairs, we suspect that the impact of L. fabarum on A. ervi,
in the experiment presented here, would have not been detectable in
the absence of the other species pair A. megourae – M. viciae.
We expected the most competitive aphid, when released from
top-down control, to have the largest impact on the community as
it is most likely to negatively affect the other species through competition. The host of L. fabarum, the aphid A. fabae, was the dominant species in the competition experiment and indeed, harvesting
L. fabarum resulted in the strongest impact on the remaining parasitoids. Furthermore, the indirect positive interactions found in our
study between the three parasitoid species followed the same direction as the competition between their hosts. These indirect positive
interactions play a major role in sustaining community persistence:
the intact community without harvesting was remarkably stable with
just two local extinctions events within 11 parasitoid generations as
compared to 22 extinctions in harvesting treatments.
Although models show that food webs are least robust to the loss
of species that have many trophic links or that occupy low trophic
levels (Eklof & Ebenman 2006; Curtsdotter et al. 2011), our experiment demonstrates that even the loss of interactions involving species that are specialised, i.e. only very few feeding links, and belong
to higher trophic levels can also lead to the extinction of indirectly
linked species from the same levels.
© 2013 Blackwell Publishing Ltd/CNRS
Letter
Although the food web structure in our experiment is not unusual for host-parasitoid networks (e.g. Vos et al. 2001), many food
webs are more complex by including specialist and generalist consumers. Our experiment has demonstrated the potential for secondary extinctions when interactions between carnivores and herbivores
are weakened, but we need further research into how food web
complexity affects the sensitivity to these effects. For example if
generalists replace the controlling effect on prey populations of a
harvested specialist consumer, they might be able to at least partially
compensate the effect and act as a buffer to secondary extinctions.
This leads us to hypothesise that networks of interacting species
become increasingly susceptible to extinctions cascades as biodiversity is lost and complexity is eroded.
We conclude that indirect positive interactions between carnivores
have the potential to maintain biodiversity in ecosystems and provide a mechanism for secondary extinctions following the loss of
carnivore species or intensive harvesting. Therefore, sustainable harvesting would need to take into account the wide-ranging ecological
impact of losing direct and indirect interactions involving the target
species rather than focusing on the management of target species’
populations only (Belgrano & Fowler 2011).
AUTHOR CONTRIBUTIONS
D.S. and F.J.F.v.V. designed the study, and wrote the manuscript.
L.S. and D.S. carried out the experiment and analysed the data.
REFERENCES
Abrams, P.A. & Nakajima, M. (2007). Does competition between resources
change the competition between their consumers to mutualism? Variations on
two themes by Vandermeer. American Naturalist, 170, 744–757.
Barnosky, A.D., Matzke, N., Tomiya, S., Wogan, G.O.U., Swartz, B., Quental,
T.B. et al. (2011). Has the Earth’s sixth mass extinction already arrived?
Nature, 471, 51–57.
Bates, D., Maechler, M. & Bolker, B. (2011). lme4: linear mixed-effects models
using S4 classes.
Belgrano, A. & Fowler, C.W. (2011). Ecosystem - Based Management for Marine
Fisheries: An Evolving Perspective. Cambridge University Press, New York.
Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R., Stevens,
M.H.H. et al. (2009). Generalized linear mixed models: a practical guide for
ecology and evolution. Trends Ecol. Evol., 24, 127–135.
Borrvall, C. & Ebenman, B. (2006). Early onset of secondary extinctions in
ecological communities following the loss of top predators. Ecol. Lett., 9, 435–
442.
Bukovinszky, T., van Veen, F.J.F., Jongema, Y. & Dicke, M. (2008). Direct and
indirect effects of resource quality on food web structure. Science, 319, 804–
807.
Curtsdotter, A., Binzer, A., Brose, U., de Castro, F., Ebenman, B., Ekl€
of, A.
et al. (2011). Robustness to secondary extinctions: comparing trait-based
sequential deletions in static and dynamic food webs. Basic Appl. Ecol., 12,
571–580.
Cushing, D.H. (1988). The Provident Sea. Cambridge Univ. Press, Cambridge, UK.
Dodson, S.I. (1970). Complementary feeding niches sustained by size-selective
predation. Limnol. Oceanogr., 15, 131–137.
Eklof, A. & Ebenman, B. (2006). Species loss and secondary extinctions in
simple and complex model communities. J. Anim. Ecol., 75, 239–246.
Gelman, A., Su, Y.-S., Yajima, M., Hill, J., Grazia Pittau, M., Kerman, J. et al.
(2012). arm: Data Analysis Using Regression and Multilevel/Hierarchical
Models.
Harrington, D.P. & Fleming, T.R. (1982). A class of rank test procedures for
censored survival data. Biometrika, 69, 553–566.
Letter
Heimpel, G.E., Neuhauser, C. & Hoogendoorn, M. (2003). Effects of parasitoid
fecundity and host resistance on indirect interactions among hosts sharing a
parasitoid. Ecol. Lett., 6, 556–566.
Heithaus, M.R., Frid, A., Wirsing, A.J. & Worm, B. (2008). Predicting
ecological consequences of marine top predator declines. Trends Ecol. Evol.,
23, 202–210.
Ives, A.R. & Cardinale, B.J. (2004). Food-web interactions govern the resistance
of communities after non-random extinctions. Nature, 429, 174–177.
Jackson, J.B.C., Kirby, M.X., Berger, W.H., Bjorndal, K.A., Botsford, L.W.,
Bourque, B.J. et al. (2001). Historical overfishing and the recent collapse of
coastal ecosystems. Science, 293, 629–638.
Janzen, D.H. (1974). Deflowering of Central-America. Nat. Hist., 83, 48–53.
McKinney, M.L. (1997). Extinction vulnerability and selectivity: combining
ecological and paleontological views. Annu. Rev. Ecol. Syst., 28, 495–516.
Meisner, M., Harmon, J. & Ives, A. (2007). Presence of an unsuitable host
diminishes the competitive superiority of an insect parasitoid: a distraction
effect. Popul. Ecol., 49, 347–355.
Memmott, J., Craze, P.G., Waser, N.M. & Price, M.V. (2007). Global warming
and the disruption of plant-pollinator interactions. Ecol. Lett., 10, 710–717.
Montoya, J.M., Pimm, S.L. & Sole, R.V. (2006). Ecological networks and their
fragility. Nature, 442, 259–264.
Myers, R.A. & Worm, B. (2003). Rapid worldwide depletion of predatory fish
communities. Nature, 423, 280–283.
Myers, R.A., Baum, J.K., Shepherd, T.D., Powers, S.P. & Peterson, C.H. (2007).
Cascading effects of the loss of apex predatory sharks from a coastal ocean.
Science, 315, 1846–1850.
Paine, R.T. (1966). Food web complexity and species diversity. American
Naturalist, 100, 65–75.
Petchey, O.L., McPhearson, P.T., Casey, T.M. & Morin, P.J. (1999).
Environmental warming alters food-web structure and ecosystem function.
Nature, 402, 69–72.
Pinheiro, J., Bates, D., DebRoy, S. & Sarkar, D. & the R Development Core
Team (2011). nlme: Linear and Nonlinear Mixed Effects Models.
Purvis, A., Gittleman, J.L., Cowlishaw, G. & Mace, G.M. (2000). Predicting
extinction risk in declining species. Proc. R. Soc. Lond. B Biol. Sci., 267, 1947–1952.
R Development Core Team (2011). R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna, Austria.
Harvesting of carnivores leads to extinction cascades 669
Sanders, D. & van Veen, F.J.F. (2012). Indirect commensalism promotes
persistence of secondary consumer species. Biol. Lett., 8, 960–963.
Schmitz, O.J., Post, E., Burns, C.E. & Johnston, K.M. (2003). Ecosystem
responses to global climate change: moving beyond color mapping. Bioscience,
53, 1199–1205.
van Veen, F.J.F. & Godfray, H.C.J. (2012). Consequences of trait changes in
host-parasitoid interactions in insect communities. In: Interaction Richness and
Complexity: Ecological and Evolutionary Aspects of Trait-Mediated Indirect Interactions
(eds. Ohgushi, T., Schmitz, O. & Holt, R.). Cambridge University Press,
Cambridge, pp. 28–46.
van Veen, F.J.F., van Holland, P.D. & Godfray, H.C.J. (2005). Stable coexistence
in insect communities due to density- and trait-mediated indirect effects.
Ecology, 86, 3182–3189.
Voigt, W., Perner, J., Davis, A.J., Eggers, T., Schumacher, J., Bahrmann, R. et al.
(2003). Trophic levels are differentially sensitive to climate. Ecology, 84, 2444–
2453.
Vos, M., Berrocal, S.M., Karamaouna, F., Hemerik, L. & Vet, L.E.M. (2001).
Plant-mediated indirect effects and the persistence of parasitoid-herbivore
communities. Ecol. Lett., 4, 38–45.
Wootton, J.T. (1994). The nature and consequences of indirect effects in
ecological communities. Annu. Rev. Ecol. Syst., 25, 443–466.
Zarnetske, P.L., Skelly, D.K. & Urban, M.C. (2012). Biotic multipliers of climate
change. Science, 336, 1516–1518.
SUPPORTING INFORMATION
Additional Supporting Information may be downloaded via the online
version of this article at Wiley Online Library (www.ecologyletters.com).
Editor, Micky Eubanks
Manuscript received 29 November 2012
First decision made 21 December 2012
Manuscript accepted 4 February 2013
© 2013 Blackwell Publishing Ltd/CNRS