Diet Breadths and Invasive Plant Impacts On Food Webs

Diet breadth influences how the impact of invasive plants is propagated through food webs
Author(s): Luisa G. Carvalheiro, Yvonne M. Buckley and Jane Memmott

Source: Ecology, Vol. 91, No. 4 (April 2010), pp. 1063-1074
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/25661148 .
Accessed: 15/04/2013 17:26
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.
http://www.jstor.org
This content downloaded from 129.24.144.237 on Mon, 15 Apr 2013 17:26:53 PM

All use subject to JSTOR Terms and Conditions
Ecology, 91(4), 2010, pp. 1063-1074

? 2010 by the Ecological Society of America
Diet
influences how the impact of invasive plants is
breadth
propagated throughfoodwebs
Luisa
3CSIRO
G.
Yvonne
Carvalheiro,1'4
M.
and
Buckley,2'3
Jane Memmott1'5
1
United Kingdom
University of Bristol, School of Biological Sciences, Woodland Road, Bristol BS81UG
4072 Australia
School of Biological
Sciences, Queensland
2University of Queensland,
Sustainable Ecosystems, Queensland Bioscience Precinct, 306 Carmody Road, St Lucia, Queensland
4067 Australia
Abstract.
Invasive
are
plants
a major
considered
cause
of
ecosystem
degradation
worldwide. While their impacts on native plants have been widely reported, there is little
informationon how these impacts propagate through food webs and affect species at higher
trophic levels. Using a quantitative food web approach we evaluated the impacts of an
invasive
plant
on
communities,
plant-herbivore-parasitoid
asking
specifically
how
diet
breadth influencesthepropagation of such impacts.Measuring the impact of the alien plant at

the plant level seriously underestimated the community-level effectof thisweed as it also
caused
changes
parasitoid
in the abundance
of native
richness.
invading
species
The
and parasitoids,
herbivores
affected
plant
specialist
along with a decrease

and generalist
subsets
in
of
communities differently,having significantand strongnegative impacts on the abundance of

consumers.
effect on generalist
specialists with no negative
to further disruptions
of top-down
regulatory mechanisms,
via
shared
natural
enemies.
invasion
Plant
competition
all
consumer
Specialist
releasing
generalist
also
significantly
decline
species
increased
led
from
the
evenness of species abundance of all trophic levels in the food webs, as well as the evenness of
interaction
species
frequency.
Extending
evaluation
impact
to higher
trophic
levels
and
considering changes in trophicdiversitywithin levels is hence essential for a full evaluation of

the consequences
of invasion
by alien
on diet breadth
information
Moreover,
plants.
of species
in the invaded community should be taken into account when evaluating/predicting the
impacts
on any
introduced
species.
alien plants; apparent competition; Avon Heath Country Park, Dorset,

Key words:
shallon; herbivore; parasitoid;
food web; Gaultheria
trophic cascade.
are
plants
studies
negative
reporting
(e.g., Minchinton
foundations
of
considered
biodiversity (Vitousek et al.
food
threat
serious
to
1997), with numerous

on
impacts
et al. 2006).
terrestrial
native
plants
Plants provide the
webs,
and
bottom-up
are a common
trophic cascades
phenomenon
(e.g., Price
Hunter
Invasion-induced
in plant
2005).
changes
an
communities
could
therefore
have
upon
impact
in other
most
levels. However,
studies
species
trophic
on the ecological
impacts of invasive plants focus solely
on the plant community
level (e.g., Hulme
and Bremner
we
find
examples
of
(Waser et al. 1996). As generalist and specialist guilds
can
react
positive effect of invasion, with the presence of alien
to community
in very distinct ways
change
and
Tscharntke
the
2007),
evaluating
on communities
of plant
invasion
in which
Rand
impacts
such as insect herbivores,
specialist
species are frequent,
reveal a different outcome
to those seen in more
may
communities.
generalized
impacts
2006, Stinson et al. 2007). Among the few studies that

evaluate the impact of introduced plants on thewider
community
et al. 2008).
(e.g., Bartomeus
systems
Plant-pollinator
are generalized
in that most plants are visited by several
and most
visit
several
pollinators
pollinators
plants
(e.g.,
and
plant-pollinator
of
received 24 November
2008; revised 26 June
7 July 2009; final version received 31 July 2009.
Editor: J. T. Cronin.
Corresponding
4 Present
address:
South African
National
Biodiversity
Research
Institute, Kirstenbosch
Centre, Claremont
7735,
South
Africa.
Cape Town,
5
author.
Corresponding
E-mail:
Jane.Memmott@bristol.ac.uk
invasive
species
Being
at a
able
to
community
predict
level
the
is a
highly desirable goal as it could guide the development

of bettermanagement strategies (Buckley et al. 2006).
To
achieve
this,
the mechanisms
that
influence
impact
propagation through natural communities need to be

understood. Despite decades of theoreticalwork linking
connectance
Manuscript
2009; accepted
diet breadth;
plants leading to an increase of flower visitor frequency
Introduction
Invasive
UK;
to stability
(e.g., MacArthur
1955, McCann
2000) and while there are indications that diet breadth

may
have
an
important
influence
(Rand
and Tscharntke
2007), its role on impact propagation remains unclear

(Fox 2007).
English
heathlands
are
anthropogenic
habitats
main
tained by grazing that,despite their low plant diversity,

provide key refugia for a high number of endangered
1063

1064
LUISA
G. CARVALHEIRO
species (Webb and Haskins 1980, Liley and Clarke

2003). The simplicityof heathland floramakes ita good
natural system inwhich to study trophic cascades (e.g.,
Forup et al. 2007). Currently heathlands in theUnited
Kingdom (UK) are threatened by invasions of intro
duced
as Gaultheria
such
plants
shallon
Pursh
(Erica
ceae) and Rhododendronponticum L. (Ericaceae) (Webb

1986). Gaultheria shallon, the subject of this study, is a
densely
growing,
shrub
evergreen
perennial,
to
native
North American heathland. This species is one of the

top 20 alien invasive species that threaten theUK flora
(Plantlife 2007).
In this studywe used a natural gradient of G. shallon
abundance to evaluate how diet breadth influences
invasive impact propagation through a plant-herbi
food
vore-parasitoid
interactions
and
of our work
objectives
sought to identifyand quantify
heathland
among
parasitoids
The
web.
were threefold. (1)We
how
evaluate
and
herbivores,
plants,
G. shallon
integrates
into
the native community. Specialist herbivores from

introduced species are usually absent from the exotic
range
have
and
a
native
generalist
greater
on
impact
are
herbivores
the
to
expected
flora
than
native
on
introduced species that have become invaders (enemy

release hypothesis, reviewed by Keane and Crawley
we predict
[2002],but see Parker et al. [2006]); therefore,
that G. shallon would have both fewer generalist and
specialist herbivores thannative plants. (2)We sought to
determine
how
G.
shallon
affects
abundance
the native
food web. As species at higher trophic levels are thought

to be at higher extinction risk than their resources
(Cronin 2004), we predict first that invasion will affect
insect
to
communities
greater
extent
than
plant
communities. (3)We sought to testwhether diet breadth

plays a key role in the bottom-up and top-down
of
propagation
more
The
impacts.
generalized
consum
ers are, themore likelytheyare to include species in their

diet that are unaffected by theweed and these can be
as
and
generalist
specialist
refugia. Moreover,
are likely to share natural
hence
enemies,
each
abundance
other's
compe
by apparent
influencing
used
consumers
1977). Therefore, we predict that diet
tition (Holt
breadth
will
impacts
on
their
consumers
how
influence
resources,
with
generalist
to
respond
consumers
being less affected than specialist consumers, and that

generalist
consumers
increasing
their densities.
that
share
enemies
natural
with
specialistswill benefit from a release of natural enemies,
ET AL.
Vol.
Ecology,
paths. The paths were mostly

with just occasional
herbaceous
jacobaea
L.
and
Taraxacum
species
can
be
associated
clear
of any
such
species
vegetation,
as Senecio
of
sp. Abundance
with
91, No.
invasive
distur
anthropogenic
bance changes (seeMacDougall

and Turkington 2005,
Didham et al. 2007). However, in our case study all
studyplots were located in a relativelyconstrained area
subjected to similarmanagement rules and to similar
environmental conditions, so the expectation is that the
G.
shallon
is a
gradient
true
invasion
gradient.
Manip
ulative experiments that add invasive species in a

controlled and replicated fashionwould provide a better
means of assessing impacts of invasive species (Siemann
those experiments
would
put already
at further risk.
such as heathlands,
habitats,
a "natural
correlational
ap
experiment"
However,
1998).
threatened
Therefore,
proach utilizing a gradient of invasion, such as the one

used in this study, is realistically the only possible
in protected
More
when
habitats.
working
approach
an advantage
of this natural
experiment
approach
over,
is that it allows system stabilization. As most experi
mental
on
studies
cascades
trophic
last one
only
or a few
resource/consumergenerations (Borer et al. 2005), they

do
not
capture
as
such
processes,
long-term
species
extinction,and may be inadequate for studyingcompo

sitional change in communities (Leibold et al. 1997).
Plot dimensions varied slightlyto avoid trails,but the
was
area
overall
surveys were
for all
equal
out every
carried
plots.
14-20
total
on
days
of
seven
each
plot
from 27 June to 15 September 2005. In each survey,one

transect (3 X 20 m) was haphazardly placed per plot,
any
cover
avoiding
percent
overlap
of each
with
plant
transects.
previous
species was
recorded,
The
and
all larval leaf herbivores were collected by beating the

onto
plants
a tray.
All larvae were individually reared until an adult

herbivore or a parasitoid emerged or the individual died.
For herbivore species found at high abundance (>20
specimens/m2), all individuals were counted but a
subsample
of insects
was
for rearing.
collected
reared
from
up.
by multiplying
these
Data
The
was
total
number
then estimated
samples
for all seven
transects
were
pooled (rearingdata) or averaged (plant abundance) for

each
plot
as
appropriate.
Herbivores
and
parasitoids
were either identifiedto the species levelormorphotyped

by
taxonomists.
Information
from
the
literature
was
sufficientto distinguish parasitoids from hyperparasi

toids. Data
were
used
to construct
quantitative
plant
herbivore-parasitoid food webs for each plot, using

softwarewritten inMATHEMATICA
were
(Wolfram Re
selected in theAvon Heath
Eight 1600-m2plots
All statistical
Illinois,
a
search,
USA).
of
lowland
site
Champaign,
composed
Country Park, Dorset, UK,
heathland dominated by Calluna vulgaris (L.) Hull (for analyses were performed with the software R (R
Core Team 2007).
grid references see Appendix: Table Al). Plots were Development
Methods
selected
in areas
in which
no
chemical
treatment
was
applied to control weeds, covering a gradient of G.

shallon abundance (0.1-80% of total plant cover), and
plots were separated by a minimum of 125 m and a
maximum
of
945
and
situated
2-5
away
from
and quantifyinginteractions:Is G.
Objective 1, identifying
shallon
attacked
less by herbivores
than are native plants?
Herbivory (measured as thenumber of individuals per

meter square of plant cover) was determined in each plot

PROPAGATION OF IMPACTS THROUGH FOOD WEBS 1065
April 2010
for each
of the abundant
plant
species,
To
normalize
i.e., those present
in all eight studyplots: G. shallon,Calluna vulgaris (L.)

Hull (heather), Erica cinerea L. (bell heather), and
Pteridium aquilinum (L.) Kuhn (bracken). General linear
mixed models (LME) were used to test differences in
herbivory between G. shallon and native plants, using
plant origin (native vs. introduced) as a fixed factor and
plot
as
random
factor.
(herbivore density) were

using
package
"nlme."
residuals,
log transformed and

errors
Gaussian
analyzed
data
tests on
and F
then
changes
simplification,using the R
in deviance during model
Objective 2: How does G. shallon abundance affect

thenativefood web?
invader, and a significant interaction between trophic

leveland specialism suggests that specieswith similardiet
breadths are affected differentlydepending on their
trophic level.To understand how diet breadth influences
the propagation of impacts we grouped species in
consumer subsets, according to diet breadth and trophic
level.
resources
The
of
consumer
each
were
species
identifiedby combining our rearingdata with published

rearing records for herbivores and for parasitoids (see
references in the Appendix). Data from the literature
enabled a more precise identificationof diet breadth, as
informationprovided by rearing data is unlikely to be
for the rare species.
sufficient
For
if a generalist
example,
species was only collected once it would be wrongly

considered a specialist. Specialistswere defined as species
Generalized linearmodels (GLM) were used to test thatfeed exclusivelyon a singlefamilyof resource species.
the effectof the abundance of G. shallon on plant and The few insects that could not be identifiedto species
were not used in this analysis (see Appendix: Tables A3
insect communities
and
richness,
(abundance,
species
evenness
and A4).
that quantifies
evenness).
Species
(a measure
consumer
were measured
resources
As
and
equality of the species within a trophic level in termsof
using
abundance) of each trophic level along with the evenness differentunits (plant abundance, percent cover; insect
of species interactions (a measure that quantifies the abundance, number of specimens), analyzing all trophic
to determine
equality of differenttrophic links in termsof frequency; levels together was not possible. Therefore,
see Tylianakis et al. 2007) were calculated for each plot
themanner inwhich the impact of G. shallon on a given
An increase
in evenness
the R package
"bipartite."
are more
to one
that species or interactions
equal
or frequency,
in terms of abundance
another
respective
using
means
were
normalize
the residuals,
data
arcsine
ly. To
or log-transformed
transformed
(evenness)
(plant per
errors or
cent cover) and then analyzed
using Gaussian
analyzed
errors
Poisson
using
and
(abundances
species
richness of insects) corrected for overdispersion (using

where
quasi-Poisson)
As
appropriate.
one
only
plot
had
consumer
subset
on
to the impact
relates
their resources,
we compared the effectof G. shallon abundance on each

consumer
subset
example,
vs.
herbivores
generalist
herbivore
on
the effect
with
on
specialists
its resources,
all
native
vs.
Ericaceae
all
for
plants;
Ericaceae
plants. Three general hypotheses can be used to provide

a conceptual
for how
framework
a decline
in their resources.
are
1.?Consumers
Hypothesis
consumers
to
respond
less negatively
affected
a G. shallon percent cover greater than 50% (79.2% G.

shallon), the leverage value of this plot was high,
potentially having a strong influenceon the statistical
models. All study plots were located in a relatively
by the invasive plant than their overall resources,
to similar management
rules,
subjected
so the expectation
is that this plot
is representative.
to explore
the robustness
of the results to
Nonetheless,
overestimate
area
constrained
the inclusion of thishigh abundance point, we repeated

the analysis using just the seven plots inwhich G. shallon
was
abundance
both
the full and
lower
than 50%.
reduced
data
Results
are
reported
for
of bottom-up
determine
whether
impacts?
generalist
and
con
sumers were affected differentlyby the invasive plant,

generalized linearmixed-effectsmodels (GLMM) were
used, using G. shallon abundance, trophic level (herbivore
vs. parasitoid), and specialism (specialistvs. generalist) as
fixed
variables
significance
and
of
the
plot
as
interactions
random
between
factor.
The
variables
was
tested using a Poisson error distribution (R package
"lme4").
significant
interaction
chain).
In
between
G.
to their resources
(i.e.,
this
on
the resource
impacts
at the consumer
level.
case,
level
impacts
are affected
2.?Consumers
to the same
Hypothesis
extent as their resource
in
species
(e.g., 50% decrease
resources
will
lead to a 50% decrease
of the consumer
species), maintaining
resources
the
(i.e.,
resource
shallon
abundance and specialism indicates that species with

differentdiet breadth are affected differentlyby the
are
level
consumer
their density relative to their
impact
of G.
shallon
is constant
a good
of
indicator
at
impacts
the
level.
Hypothesis
specialist
relative
going up the food web). In this case, impacts at the
analyses.
Objective 3: Does diet breadth influencethepropagation

To
their density
increasing
the impact of G. shallon diminishes furtherup the food
3.?Consumer
species
are more
negatively
affected than their resource species, decreasing their

density
relative
to their resources
(i.e.,
the impact
of G.
shallon ismagnified going up the foodweb). In thiscase,

on the resource
impacts
the consumer
level.
level underestimate
impacts
at
To identifywhich consumer subsets conform with

which hypothesis,we used GLM to test the effectof G.
shallon
resources.
abundance
on
consumer
When
subsets
to normalize
necessary,
on plant abundance
were
log-transformed
count data
errors, while
using Gaussian

and
on
their
data
residuals,
and analyzed
were
analyzed
1066
LUISA
with
error.
Poisson
where
G. CARVALHEIRO
was
accounted
Overdispersion
As
native
could
communities
appropriate.
for
re
spond differentlyto different levels of invasion,models

were
tested
also
for quadratic
effects of G.
cover.
shallon
To testwhich model (linear vs. quadratic) best fitted the

the most
data,
correction
second-order
was
model
parsimonious
as
selected
that
information criterion with a
with the lowest Akaike
for
small
sizes,
sample
AICC
(Burnham and Anderson 2002) and with quasi-AICc

values used for overdispersed models (Richards 2008).
As the number of data points is low (n= 8) and, hence,
our
selection
model
low
had
procedure
for
power
detecting the more detailed patterns of a quadratic

model, borderline significance (0.05 < P < 0.09) was
when
considered
the results.
discussing
To testhow indirect interactionsbetween specieswith

differentdiet breadths were affectedby the invasion,we
evaluated the potential for apparent competition (via
shared
herbivores.
between
parasitoids)
For
each
plot,
we firstcalculated the apparent competition coefficients

(dy,
the fraction
i.e.,
the parasitoids
of
one
attacking
herbivore species, /,that are likelyto have developed on

another species j; seeMuller et al. 1999) for each pair of
herbivores and then tested the effectof G. shallon on the
average
specialist-generalist
and
dy,
vice
versa,
using
error distribution. If this

GLM with a Gaussian
coefficient is high, then the indirecteffectof host j on i
is likely to be great. We then tested the effect of
on
rates
parasitism
a Poisson
with
error
where
dispersion
consumer
using GLM
for over
corrected
abundance,
distribution
As
appropriate.
abundance
parasitoid
shallon abundance
is indirectly affected by G.
to statistically
it is important
separate
and parasitoid
abundance
G. shallon
herbivores,
effect of both
If G.
herbivores.
abundance
parasitoid
may
including
effect of
this variable
complete
and
the
be
also
even
alone,
variable.
this
the
on
affects herbivore
significantly
with
abundance,
parasitoid
shallon
is colinear
but
subsets,
via
in models
significant
if there is no causative
if colinearity
is not
is
abundance
parasitoid
However,
effect
of
strong enough, a significanteffectof thisvariable will be

over
detected
and
the effect
above
are
when both variables

Therefore,
we
always
inmodels
abundance
these
represent
herbivore
evaluated
including
P
significant
models,
of G.
shallon
(i.e.,
included in the model).

G.
the effect of parasitoid

In
shallon abundance.
values
for
parasitoids
an
on
shared
the effect
of
parasitoids
the effect of G. shallon
subset over and above
on
abundance
herbivores.
All
analyses
were
repeated
using just the seven plots inwhich G. shallon abundance

was
lower
than
50%.
species.
Three
orders
of herbivores
were
collected:
Coleoptera (one species) comprised 87.4% of specimens,

Lepidoptera (26 species) comprised 11.0% of specimens,
and Hymenoptera (one species of Symphyta) comprised
1.6% of specimens. A total of 2172 parasitoids were

collected,which consisted of 15 species of Hymenoptera
and two species of Diptera. Detailed information on
community species composition is provided inAppen
dix: Tables A2-A4. Finally, three specimens of a
generalist obligatory endophagous hyperparasitoid spe
cies
sp.) were
(Mesochorus
Gymnoscelis
from the Lepidopteran

As itwas not possible
reared
Haworth.
rufifasciata
to identify
which parasitoid species of theG. rufifasciata
were used as hosts by the hyperparasitoid, thiswas not
included in the food web data set. Fig. 1 illustrates the
impact of 0.4%, 30.0%, and 79.2% of G. shallon on
heathland foodwebs (all eight foodwebs are provided in
theAppendix: Fig. Al).
and quantifying interactions:
Is G.
less by herbivores
attacked
shallon
The vast majority (99.8%) of herbivore rearings came

from native plants, and only 10 specimens (six species/
morphospecies)
with
compared
were
and
generalist
vores,
FU23
herbivores/m2,
specialist
shallon
0.0
When
plant
species,
significantly less by both
0.04,
native
F123
shallon.
native
herbivores
specialist
P <
4.8,
herbivores,
common
attacked
mean
G.
from
collected
the most
shallon was
G.
G.
0.5
plants
herbivores/m2,
5.7,
mean
herbi
(generalist
mean
0.1
herbivores/m2;
G.
0.03, mean
<
shallon
native
plants
3.2
herbivores/m2).
Objective 2: How does G. shallon abundance affect

thenativefood web?
The
statistics described in this section for plant,
are presented
in
and parasitoid
communities
herbivore,
on native
1. Although
the impact of G. shallon
was
unclear
abundance
(significant
community
plant
Table
negative effectswere not detectable below 50% level of

invasion),
significant
abundance
community
effects
(both
were
insect
for
detected
and
herbivores
parasit
oids) in both the full (all eight plots) and the reduced
(<50%) data set.Although plants and herbivore species
richness
were
not
significantly
by G.
affected
shallon
invasion, a significantdecline was detected on the full

data
set of parasitoid
invasion
significantly
species richness. Gaultheria

the evenness
increased
shallon
of
all
trophic levels (full and reduced data sets), as well as

evenness
interactions
of species
=
evenness
F] 6
[IE],
=
7.5,
IE, FU6
herbivore-parasitoid
the
interaction
(plant-herbivore
<
P
0.05;
7.2,
P
<
0.04),
effect being more evident below 50% weed
parasitoid
total of 3603 herbivores were collected from 12
91, No.
Vol.
Ecology,
=
12.5, P <
IE, Fl 5
=
IE, FU5
9.1, P < 0.03).
(plant-herbivore
Results
plant
ET AL.
0.02;
this
invasion
herbivore

of bottom-up
impacts?
The statisticsdescribed below are presented inFigs. 2

and 3 with model details provided in Table 2. The

April 2010
Ill
Herbivores
. , ?.
16 4
; }-l
Plants . ^^^^
Parasitoids
Parasitoids
4 5 10151923^
1 2
10 11
5 16
6
14
Herbivores
12
10
11
12
1 13
Parasitoids
J parasitoids I-1
'
Herbivores 1
100
|?i
herbivores'-'
s^Jgg
Plants ^^^^^^^^H^^^^^^^^^H
HHHHHHHHI^Hii^^^^H^^^^Î
\
1
Ljûj
7 9
10
12
10%plant i-1
coverage I_I
I
I
11
Fig. 1. Plant-herbivore-parasitoid
food webs for three plots representative of the pattern found along the Gaultheria shallon
invasion gradient for all eight plots: (a) 0.4%, (b) 30.0%, (c) 79.2% G. shallon. Each species is represented by a rectangle, the widths
to their abundance
at the field site. The size of the triangles connecting
them represents the frequency of
being proportional
lines represent interactions reported in the literature, which were not detected during the
interactions in the study area. Dotted
sampling surveys. Invasive plants (G. shallon and the rarer Rhododendron ponticum) are shown in black as are generalist consumers
For explanations
of species codes, see Appendix:
Table A2. The study was conducted
in the Avon
(herbivores and parasitoids).
Heath
Country
statistics
using
Park, Dorset,
UK.
the reduced
data
set (only
presented inAppendix: Table A5.

Diet
were
breadth
affected
significantly
by G.
shallon
are
how
with
consumers
specialist
species being negatively affected by the invasion,while

generalist
consumers
were
not always
(full and
reduced
negatively
affected
generalist
data
factor
abundance]:
influenced
abundance,
seven plots)
sets,
interaction
P <
[specialism],
consumers
belonging
factor
0.0001).
to
[G. shallon
Moreover,
different
trophic
levelswere affecteddifferently(fulldata set, interaction

factor [specialism]: factor [trophic level], P < 0.0001;
reduced
data
set,
interaction

factor
[specialism]:
factor
1068
LUISA
Table
1.
Impact
of Gaultheria
ET AL.
G. CARVALHEIRO
shallon on native communities
of plants, herbivores,
and parasitoids.
All plots
Response variableby community
91, No.
Vol.
Ecology,
Plots
P
df
below
invasion
50%
df P
Plants
Abundance
(log transformed, Gaussian
Evenness
(arcsine
-0.0133
error)
Species richness(Poisson error)
transformed, Gaussian
0.0067
error)
NS6 5
<0.003
NS6
5 NS
<0.002
5
<0.0001
Herbivores
Abundance
error)
(quasi-Poisson
NS 6 NS 5
Species richness (Poisson error)
Evenness
(arcsine transformed, Gaussian
error)
-0.0491
<0.002
<0.004
5
0.0072
<0.006
<0.025
-0.0457
5NS 6
0.0038
<0.009
5
<0.03
<0.003
<0.045
Parasitoids
Abundance
error)
(quasi-Poisson
?
Species richness (Poisson error)
Evenness
(arcsine transformed, Gaussian
<0.04
error)
Notes: Values
had a
indicating the strength of G. shallon impact (p value) are presented whenever G. shallon abundance
significant effect. P values presented were obtained from likelihood ratio tests comparing deviances with and without G. shallon
in themodel with all eight plots as well as in themodel with only the seven plots below 50% invasion. The abbreviation
abundance
in the Avon Heath Country Park, Dorset, UK.
NS
indicates P > 0.05. The study was conducted
[trophic level],P < 0.05), with generalist parasitoids not

being significantlyaffected,while generalist herbivores
either of the two exotic Ericaceae present in the study

area
(G. shallon
were.
were
considered
Using thedefinitionof a specialist being a species that

feeds on resource species belonging to a single family
and combining the rearing data with informationfrom
the literature,six groups of consumers were identified:
(1) generalist herbivores (16 species); (2) generalist
parasitoids on Lepidoptera (seven species); (3) herbi
vores specialist on Ericaceae plants (nine species); (4)
parasitoids specialist on Chrysomelidae (Coleoptera)
(one species); (5) parasitoids specialist on Noctuidae
(Lepidoptera) (three species); and (6) parasitoids spe
cialist on Geometridae (Lepidoptera) (one species). As
Gaultheria shallon abundance affected specialist and

generalist subsets of communities differently,having
significantand strong negative impacts on the abun
dance of all specialist consumer subsets (herbivores and
parasitoids; Fig. 3b, d, f,h), while no clear negative
effectwas detected for generalist consumer subsets (Fig.
2b, d). The six consumer subsets responded toG. shallon
specialist
never
were
on Ericaceae
herbivores
found
1. Generalist
their resources
shallon,
while
II l^^*^
II 60=
<
^
To 30
R ? CD
<
.q
40-
levels
of
invasion
unaffected
were
by G.
positively
Generalist consumers
Herbivores
b)
>v Q
10 Ho
2s
_(_
~~\?
d) Parasitoids
"d
30-Lj_j_!_
c) Lepidoptera
7nJ?
"d
low
(i.e., native plants) were

herbivores
the generalist
Ij
?>?\
JI20:
/o?
CD
herbivores.?At
affected.At higher levels of the invasion both generalist
on
-j0a) Native plants
only native Ericaceae

this consumer
subset.
as follows.
Resources
-c
and R. ponticum),
as resources
for
_ ^s.
n-1-1-1?=lr
20
0
,q~
^
40
60
80
n-1-1-1-t*
20
40
60
80
G. shallon cover (%)

Effect of Gaultheria shallon on the abundance of generalist consumer subsets and their resources. Regression
are presented whenever
the effect of G. shallon on the response
that best fit the data (linear vs. quadratic)
in Table 2.
details are provided
significant. Model
Fig.
2.
models

lines ofthe
is
variable
April 2010
Resources
8?
fl a) Ericaceae
Rn
?^S2
c
60 -
eg
E?
11
C "D
3-?
-\
400-
5-E
\o
?'i >
_io
40
20-
"c
7kt5
~
ci
?-h-1-1-i^t
S?
co
D.^
eV
0\
0 I,>-4h-,-*
1-m o o
^-jg
=jo3
o
f)Parasitoids
of Geometridae
<2
-1
0
h) Parasitoids
ofNoctuidae
1-
^.E
>-+
?Qi
3H
8? o
^ ^
"D>
40
"\
400
g)Noctuidae
n-1-1-1-R-
20
-D
< 6
^n,^^^
^^H^^
d) Parasitoids
of Chrysomelidae
3 1000-
"
gVp
Xa 15-
"V
^3 o 800H<>
o.
li
e) Geometridae
_?^?s.
400-
5-.
*|-1-*
^J*
c t5
b) Herbivoresof Ericaceae
H
? =-g 1000 v
HOc) Chrysomelidae
'|T
<
^5^^
Specialist consumers
^Hh.
^^^^
^ ^
5^%hhhn,s
^ c!ci ?~n-1?QiLi?9
60
80
40
20
60
80
G. shallon cover (%)

of specialist consumer subsets and their resources. Regression
3. Effect of Gaultheria shallon on the abundance
are presented whenever
that best fit the data (linear vs. quadratic)
the effect of G. shallon on the response
details are provided
in Table 2.
significant. Model
Fig.
models
herbivores
(Fig.
their resources
and
2a, b). The
were negatively
affected
effect on generalist
herbivores
positive
was significantin both the full (all plots analyzed) and

reduced (seven plots) data sets,while the negative effect
was only significant in the reduced data set for the
consumer
an
subset.
in
increase
resources,
consumers
Therefore,
consumer
although
combined
2. Generalist
H\,
supporting
relative
density
this was
with
there was
to
their
to a positive
a neutral
effect on
effect on
resources.
their
resources
due
parasitoids.?Although
(Lepidoptera) were negatively affectedby G. shallon (in

both fulland reduced data sets),no significanteffectwas
detected on the abundance of thisconsumer subset (Fig.
2c, d). Therefore, supportingH\, generalist parasitoids
were
less
increase
affected
than
in their density
their
resources,
leading
to their resources.
relative
3. Specialist
herbivores.?At
resources
of this consumer
low levels of invasion

subset
(i.e., Ericaceae)
to an
the
were
not significantlyaffected by G. shallon,while specialist

herbivoreswere negatively affected (Fig. 3a, b). Only at
the highest invasion level did resources decline, while
consumers
were
also
were
not
affected.
at
H3,
supporting
shallon
abundance
herbivore
density
Gaultheria
in the reduced
detected
data
to moderate
low
to
led
to their resources,
relative
shallon
effects
set. Therefore,
levels
invasion
reduction
lines of the
variable
is
of
G.
specialist
i.e., impacts
on specialistherbivoreswere not easily predictable based

on
impacts on their resources.

con
4. Chrysomelidae
specialist
parasitoids.?Both
sumer and resource
subsets were negatively
affected by
G.
shallon
at
abundance
low
levels
of
invasion
(Fig.
3c, d), and inboth full and reduced data sets.Therefore,

supporting
H2,
resources
and
consumers
were
equally
affected; consumersmaintained theirdensity relative to

their resources.
5. Geometridae
specialist
parasitoids.?At
low
to
moderate levels of invasion the resources of this

consumer subset were not significantlyaffected by G.
shallon,while these specialist parasitoids were negatively
affected
(Fig.
3e, f).
In
the reduced
data
set, resources
were not significantlyaffected, while this consumer

subset was still negatively affected.These data suggest

1070
LUISA
Table
2. Consumer/resource
models:
comparison
resources using the full data set (eight plots).
(R) and
(C) pair
Resource
consumer
G. shallon
G. CARVALHEIRO
ET AL.
of linear and quadratic
Vol.
Ecology,
models
for each consumer
obtained
91, No.
subset and
their
??
AICC Best
(G. shallon)2
model
equation
Generalists
R: Native
Model
error)
(log-transformed data, Gaussian
<0.001
NS
Model
1
1
<0.003
plants
2
C: Generalist
Model
(no data transformation, Poisson

<0.001
<0.0002
Model
1
81
NS
herbivores
R: Lepidoptera
error)
transformation, Poisson error)

<0.002
NS
Model
1
67
<0.0001
(no data
2
Model
C: Generalist
-12
(no data
Model
NS 1
parasitoids
transformation, Poisson
35no effect
59
ln(C)
60
\n(R)
1)
= 4.526 +
0.0038G
2.6203 +
0.0359G
4.1768 +
0.0007G
0.0002G2
0.0005G2
0.0004G2
error)
40
NS NS
2Model
ln(R +
Specialists
data, Gaussian
error)
<0.001
NS
Model
1
1
<0.007
R: Ericaceae
2
Model
C:
(log-transformed
herbivores
Specialist
Model
R: Chrysomelidae
Model
(no data
C: Specialist
Model
Model
parasitoids
transformation, quasi-Poisson
NS
<0.02
1 Model
12
<0.02
on Geometridae
<0.03
1Model
<0.02
Model 2
(no data
R: Noctuidae
transformation,
<0.0002
Model
1
C: Specialist
parasitoids
error)
error)
13
error)
-
on Noctuidae
(no data
<0.04
1 Model
28
1)
6.917
6.863
\n(R)
69
\n(R)
ln(#)
47
transformation, Gaussian
26
error)
C=
NS
0.0003
X G2
0.0729
X G +
0.0004
X G2
1.9384
X G2
0.0560
X G +
0.0002
X G2
0.0210
X G -
0.0008
X G2
X G
0.0159
3.2241
0.0002
X G +
X G -
0.0662
3.2134 +
error)
C = 0.8998
0.0002
4.2874 +
error)
=
ln(C)
10
NS
50
<0.06
Model 2
quasi-Poisson
10
(no data- transformation, Gaussian

13
Poisson
NS NS
2Model
\n(R +
transformation, quasi-Poisson
error)
= 6.3631
12
ln(C)
transformed data, Poisson

<0.003
NS
Model
1
79
<0.001
(no data
Specialist
transformation,
NS
on Chrysomelidae
(no data
NS
<0.07
1 Model
15
<0.009
parasitoids
R: Geometridae
C:
on Ericaceae
(no data
<0.002
1Model
12
<0.002
-13
X G
0.0174
X G
0.0324
Notes: Count data were analyzed using Poisson distribution errors, and percent cover data were log transformed and analyzed
variables used were:
with Gaussian
distribution errors. Best model equations are always based on untransformed data. Explanatory
P values were obtained from a likelihood ratio test in
Gaultheria shallon abundance
(G) and the square of G. shallon abundance.
indicates P > 0.05; for quadratic
which deviances with and without that term in themodel were compared. For linear models, NS
a significant effect of G. shallon abundance
(or its square) was detected, the best model
models, NS indicates P > 0.09. Whenever
is provided for each subset. An ellipsis indicates a variable that was not included in the model.
(lowest AICC)
that,
supporting
G.
H3,
shallon
to
led
abundance
reduction of specialist herbivore density relative to their

resources,
i.e.,
impacts
on
this
specialist
parasitoid
subsetwere not easily predictable based on impacts on

their resources.
6. Noctuidae
and
resource
specialist
parasitoids.?Both
were
negatively
subsets
consumer
affected
shallon abundance at low levels of invasion (Fig. 3g, h),

although
in the reduced
data
sets, only
consumers
were
still negatively affected. These results suggest support
for H3,
G.
shallon
abundance
led
to a
reduction
i.e.,
of
on
impacts
density
this
relative
specialist
not easily predictable based
to their
on
resources,
subset
parasitoid
impacts on
were
their
resources.
regard to indirect food web
With
shallon
by G.
herbivore
specialist
had
significant
negative
interactions, G.
effect on
the average
specialist-generalist dy (apparent competition strength;

see Appendix: Fig. A2) on both full and reduced data
sets (Table 3 and Appendix A: Table A6), revealing that
generalist
herbivores
regulating
effects.
The
were
released
inclusion

from
of parasitism
top-down
rate in
April 2010
3.
Table
Apparent
tests using
competition
the full data
Response variable (Y)

dy,
herbivores
Generalist-specialist
error distribution
Gaussian
dy,
herbivores
abundance,
0.0606
0.0037 G
<ô.uô
ain<rn nss
+ 0.000002 G2
=
Y
0.0096
0.000003 G
N^n
UU0 3iy
+ 0.00001 G2
?
In (Y)
2.8177 + 0.0368 G
^cxacscw
0.0004 G
3.51 \p
= 5.5792 + 0.0918 G
ln(y)
^-n nn?
ûuuz
quasi-Poissonerrordistribution
Residual
Parasitism
(G.
shallon
rate (p)
shallon)2
Y =
herbivores abundance,
error distribution
Specialist
G.
Model
herbivores
Specialist-generalist
error distribution
Gaussian
Generalist
Poisson
set (eight plots).
df
AICC
cvrc
_9Q
n.
_,Q
s
- 0.0008G2 +
2.3557/7
/nmm
^nnnm
<^nm
û'u^
<-n m
ûuj
a ao
4n1j
Notes: Explanatory
variables used for apparent competition coefficient (dy) models were: Gaultheria shallon abundance
(G) and
the square of G. shallon abundance
(G2). For herbivore abundance models parasitism rate (p) was included. P values obtained from
a likelihood ratio test are presented. For linear models: NS indicates P > 0.05; for quadratic models, NS
indicates P > 0.09. The
most parsimonious
information criterion, AICC) were selected (linear vs. quadratic model). The AICC
(lowest Akaike
dy models
values of herbivore abundance models
including the variable parasitism rate are provided for comparison with models without
parasitism rate (Table 2).
the generalist herbivore abundance model led to an

increase of themodel fit(seeAICC values inTables 2 and
3), abundance being significantlynegatively affectedby
parasitism over and above the effectof the invasiveplant
for both full and reduced data sets (Table 3 and
Appendix A: Table A6). Although in the full data set
a positive effectwas detected on generalist-specialist dy,
specialist herbivore abundance models fit did not
with
improve
the inclusion
of parasitism
rate.
due to the plant's defensive chemicals that are lethal to

nonadapted native herbivores (Wiklund 1975, Chew
1977), which may also reduce pathogen attack and/or
enable allelopathy (Carpenter and Cappuccino 2005).
2: How
Objective
The
well
does G.
abundance
shallon
affect thenativefood web?
importance
recognized
of bottom-up
(e.g., Balciunas
in ecosystems
effects
and
Lawler
is
1995, Price
and Hunter 2005). The work here presented provides a

furtherexample and is one of thefirstto use a food web
Discussion
Most studies of invasive plants evaluate their impact

on
to examine
of an invasive
approach
plant
impacts
at just theplant level,with littlebeing known about the herbivores and
et
also
Heleno
al.
parasitoids (see
2009).
propagation of these impacts through the associated While quantitative foodwebs are a veryuseful toolwhen
network. We
have
that mea
demonstrated
ecological
trying to understand the complexity of interactions at
suring the impact of the invasive G. shallon at just the the community level, theydo have some limitations.We
plant level seriouslyunderestimates theoverall impactof could have had differentialsuccess in rearingparasitoid
the weed at the community level, which not only
of species abundance,
the estimates
species,
influencing
propagated upwards in the foodweb (bottom-up effects) and by removing the herbivores from the field for
but
to further top-down
effects, via apparent
in some consumer
Impacts were magnified
in food web
structure.
In this
leading to changes
also
led
competition.
subsets,
section,
we
consider
the results
obtained
in light of our
original predictions, while discussing the limitations,

highlighting the implications of our results for conser
vation
and considering
the role of food webs as
biology,
tools
when
the impact
of alien
predictive
assessing
organisms.
and quantifyinginteractions:Is G.
shallon
As
plants
shallon
attacked
seen
(e.g.,
was
less by herbivores
in other
of native
comparisons
and
Carpenter
Cappuccino
less attacked
by herbivores
and
exotic
G.
2005),
than were
native plants. As the specialist herbivore species studied

here feed on the familyof resources towhich G. shallon
belongs
(Ericaceae),
host
switching
could
have
oc
curred (Keane and Crawley 2002). However, it is clear

that forUK heathland specialist herbivores and for the
of generalist
G. shallon
is not
herbivores,
majority
resource.
alternative
This
low level of herbivory may
we may
rearing,
have
some
prevented
attack.
parasitoid
Nevertheless, these limitationsapply equally to all plots

and
across
and
are,
all parasitoid-host
nearly
sampling
design.
As predicted,
both
nities
were
more
food web
to
unlikely
consequently,
herbivore
and
affected
than
studies
confound
the
commu
parasitoid
the native
plant
community. It is possible that the effect detected at

higher trophic levels is due to a combination of loss of
resource
combined
with
in host
changes
plant
quality,
trophic
interac
due to competitionwith invasiveplants (Bukovinszky et

al. 2008). The invasion also led to an increase of
evenness
within
tions. Although
environmental
by Morozov
evenness
levels
trophic
of
and
changes of species evenness with
perturbations
can
be
diverse
(reviewed
and Li 2008), higher values of species
are commonly
associated
with
higher
values
of
species richness (Stirling and Wilsey 2001). Previous

studies inwhich habitatmodification led to a decrease of
evenness
interaction
an
generalist
be
decrease
on
dominated
systems
by
et al. 2007),
the
(e.g., Tylianakis
evenness
interaction
most
being
likely
focus
consumers
of

1072
LUISA
to
related
the
loss
resource
of
G.
and
richness
species
consequent intensificationof interactionwith the few

resources
in our
left. However,
case
webs
food
study,
were dominated by specialist species that have a high

frequency of trophic interactions with their resource
species, and hence undisturbed food webs have low
interaction evenness. By mainly affecting abundant
specialist species, the invasion led to a reduction of the
importance of their trophic links, leading to an increase
in the evenness
of species
the
Moreover,
frequencies.
and
of interaction
evenness
values
abundances
low
no known
external
factor
the invasion.
drives
were
to the same
affected
Vol.
Ecology,
as
extent
91, No.
their resources
(H2),
others (specialist herbivores and parasitoids of Geo

more affect
metridae and Noctuidae) were significantly
ed than their resources, declining in density relative to
resources
results
(Hy). These
cases
that in certain
suggest
thedecline in resources affected the reproductive success

of specialist species (i.e., the parametric threshold
[Bascompte 2003] was met), and hence their decline
was
more
accentuated
than expected.
Implications of the resultsfor usingfood webs
in
noninvaded plots might be a result of the simplicityof

this habitat, which may be associated with a high
number of available niches, potentially increasing
invasibility (reviewed by Hillebrand et al. 2008), even
when
ET AL.
CARVALHEIRO
as predictive
tools
Extinction simulation studies of food webs thataim to

the
predict
local
of
consequences
community-level
species extinction (Dunne et al. 2002, Memmott et al.

2004) assume that as long as one resource species is still
available,
consumer
will
species
remain
in the food web.

of bottom up impacts?
For
Similar to Rand and Tscharntke (2007), who com

pared groups of consumers with differentdiet breadth
above the parametric threshold). The results of this

work indicate that a reduction in abundance of a
(generalist
vs.
predators
to
responses
contrasting
In
invasion.
case
the
resource
of G.
as some
shallon invasion these differenceswere detected within

same
the
of
group
generalist
vs.
(i.e.,
specialist
vs. generalist
para
consumers
and
herbivores
specialist
consumers
increased
to
relative
in density
their resources (Hi). If only some of the resource

species of a given consumer are affected by the invader
are
all resources
assuming
insect
further
support
foraging),
and
unsaturated
can
(i.e.,
the consumer
be
may
able to adapt its diet and use the unaffected resources

thereby
refugia,
herbivores,
generalist
the non-Ericaceae
plants
can
resources
as well
For
abundance.
maintaining
unaffected
as plants
include
in
occurring
for generalist
habitats;
parasitoids,
surrounding
of
resources
the generalist
include
unaffected
species
the
they feed. The
the herbivore family upon which
herbivores
of generalist
in abundance
increase
detected
in early stages of invasion (Fig. 2b) suggests, however,

that
this group
of consumers
et al.
Morris
miner
community,
a further
showed
(2004)
two
experimentally
had
advantage.
that by removing
of herbivores
species
other herbivore
species
from
that
leaf
shared
natural enemies with the removed species had lower

parasitism
herbivore
rates
and
abundances.
consequent
increase
of
In agreement
with Morris
parasitoids
released
generalist
herbivores
All
consumer
negatively
subsets
affected
on
effect
consumer
subsets
suffered
(parasitoids)
declines
in species richness, even when the species richness of

theirresourceswas not affectedby the alien plant (Table
1). These results suggest that local extinction of insects
occur
network.
before
Thus,
all
species are lost from the

simulations
may provide
resource
these
although
accurate predictions for systems in which generalist

species are the most frequent (e.g., plant pollination
food webs), theymay underestimate impacts on more
the mechanisms
is needed
on
Based
improved.
the one
as
such
communities
specialized
Understanding
tion of impacts
that
if these
simulations
results
presented
the
and
(herbivores
shallon.
G.
by
However, while some (parasitoids of Chrysomelidae)
here.
studied
influence
propaga
are to be
here,
we
predict that ifonly part of the resources are affected, the

consumer
will
go extinct
if all resources
only
are
lost (as
previous extinction simulation studies predict); but ifall

resources
of
consumer
consumer
are
species will go
lost from the network.
decline
extinct
the
in abundance,
their resources
before
Information on diet breadth of the species may

considerably improve thepredictivepowers of foodwebs.
environmental
example,
interaction
likely to increase
For
perturbations
evenness
when
are
more
the system
is dominated by specialist species (as in this study), but

interaction
evenness
if the system
is dominated
by generalist species (e.g., Tylianakis et al. 2007).
from
of
abundance
the
et al.
of their abundance.
specialist
were
parasitoids)
an
elicit
decrease
top-down regulating forces (Table 3), leading to the

increase
can
remain
(i.e.,
populations
those
(2004), our apparent competition analysis revealed

that a decline of an abundant herbivore subset
(specialist herbivores) and the consequent decline of
their
consumer
to sustain
specialist consumers, possibly leading to species loss,
could
sitic wasps).
Generalist
as
sufficient
study
our
wasps),
parasitic
that differences in diet breadth generate
detected
this to happen the remaining resources must be
Implications
of the results for
ecosystem
conservation
Previous studies suggest thathigher trophic levelsmay

provide a better proxy for estimating the health of an
ecosystem than the lower levels (reviewedbyHudson et al.
2006). However, thisstudy indicates that it is importantto
consider the specialization levelof these species if theyare
to be used as bio-indicators, as generalist speciesmight
indicate that impacts are attenuated, while specialist

April 2010
speciesmay indicatemagnification of impacts.The results

of our work show that the impactsof invasiveplants can
propagate up throughthefoodweb, leading to changes in
communitystructureand loss of trophicdiversity(i.e., loss
of specialist species). The implications of the loss of
trophic diversity can be profound as specialist natural
enemies
abundance
than
role on resource
regulating
natural
enemies
and
(Hassell
a stronger
have
may
generalist
May 1986). Thus, generalization of invaded communities

may disrupt top-down forces,whichmay lead to herbivore
In fact, our
outbreaks.
species
that bottom
study revealed
up cascade of effectscaused by an invasion can lead to

further
recent
with
via
effects,
top-down
Together
studies
apparent
competition.
et al.
Carvalheiro
(e.g.,
2008) thatrevealed that safetyof highly specificbiocontrol

agentsmay be compromisedby apparent competition,our
resultshighlight the importanceof taking into consider
ation the complexity of food web interactionswhen
evaluating or predicting the negative impacts of any
introduced species (either weed or biocontrol agent),
assessing
on both
changes
direct
indirect
and
interactions.
Finally, as disturbance by invasive species is leading
to a more
generalized
and
community,
as
some
previous
studies have shown that higher connectance in food

webs
increases
to species
their robustness
loss
et
(Dunne
al. 2002), disturbed communities may become more

robust
to
important
future
result
This
disruptions.
implications
in habitat
have
may
restoration
as,
also
pointed out by Tylianakis (2008), it suggests that

disturbed generalized communities could be resistant
to attempts to restore the original trophically diverse
ecosystems.
This studyhighlights the fragilityof highly specialized

communities
to environmental
such
perturbations,
as
plant invasion, providing new insights into themecha
nisms
that
influence food webs

and revealing
important
in community
structure. Given
that all ecology
in a community
to make
able
context,
being
of
to
predictions
consequences
community-level
disrup
changes
occurs
tionwould be a major advance. Our findings indicate

thatby integratinginformationon both diet breadth and
the proportion
that evaluate
of affected
consequences
resources,
of
may be significantlyimproved.
predictive
ecosystems
studies
disruptions
Acknowledgments
We
thank T. Branston and I. Cross
(Avon Heath Country
for field sites access and
Park)
information; H. Kirk, E.
R. Torres, and R. Gibson
for field and laboratory
Barbosa,
assistance; M. Bailey, J.Deeming, G. Broad, M. Shaw, and A.
Polascek
for insect identification and information on parasitoid
host range; and M. Crawley, S. Hiscock,

S. Pearce, M. Devoto,
R. Heleno,
and four anonymous
referees for
Pocock,
comments on themanuscript. L. G. Carvalheiro was funded
by
e Tecnologia
Fundacao
Y. M.
para a Ciencia
(Portugal).
was
an
funded
Australian
Research
Council
Buckley
by
Australian Research
Fellowship
(DP0771387).
M.
Cited
and S. P. Lawler.
1995. Effects of basal
Balciunas,
D.,
food
resources, predation, and alternative prey inmicrocosm
chains. Ecology
76:1327-1336.
2008. Contrasting
Bartomeus,
I., M. Vila, and L. Santamaria.
net. Oecologia
effects of invasive plants in plant-pollinator
155:761-770.
J. 2003. Extinction thresholds: insights from simple

Bascompte,
models. Annales Zoologici
Fennici 40:99-114.
J. B. Shurin, K. E. Anderson,
Borer, E. T., E. W. Seabloom,
C. A.
Halpern.
cascade?
B.
Blanchette,
2005. What
Broitman,
determines
and B. S.
S. D. Cooper,
the strength of a trophic
86:528-537.
Ecology
et al. 2006. Management
of plant invasions
Buckley, Y. M.,
mediated
interactions.
Journal of Applied
by frugivore
Ecology 43:848-857.
T., F. J. F. van Veen, Y.
Bukovinszky,
Jongema, and M.
Dickel.
2008. Direct and indirect effects of resource quality
on food web structure. Science 319:804-807.

2002. Model
selection
Burnham, K. P., and D. R. Anderson.
and multi-model
inference: a practical
information-theoretic
Second edition. Springer, New York, New York,
approach.
USA.
2005. Herbivory,
time since
Carpenter, D., and N. Cappuccino.
introduction and the invasiveness of exotic plants. Journal of

Ecology 93:315-321.
L. G., Y. M.
R. Ventim,
and J.
Carvalheiro,
Buckley,
Memmott.
2008. Apparent
the
competition can compromise
safety of highly specific biocontrol
agents. Ecology Letters
11:690-700.
Chew, F. S.
cruciferous
1977. Coevolution
of pierid butterflies and their

of eggs on potential
foodplants. 2. Distribution
Evolution
31:568-579.
foodplants.
extinction and colonization
Cronin, J. T. 2004. Host-parasitoid
in a fragmented prairie landscape. Oecologia
139:503-514.
R. K., J. M. Tylianakis, M. A. Hutchison,
R. M.
Didham,
2007. Are invasive species the
Ewers, and N. J. Gemmell.
drivers of ecological
in Ecology
Trends
and
change?
Evolution
22:489-496.
J. A., R. J. Williams,
and N.
Dunne,
Network
structure and biodiversity
remarks
Concluding
Literature
robustness
558-567.
increases with
connectance.
D.
loss
Martinez.
2002.
in food
Ecology
webs:
Letters 5:
Forup, M. L., K. S. E. Henson, P. G. Craze, and J.Memmott.

2007. The
of ecological
restoration
interactions:
plant
on ancient and restored heathlands.
networks
pollinator
Journal of Applied Ecology 45:742-752.
Fox, J.W. 2007. The dynamics of top-down and bottom-up
effects in food webs of varying prey diversity, composition,
and productivity. Oikos
116:189-200.
1986. Generalist
and specialist
Hassell, M. P., and R. M. May.
natural enemies in insect predator-prey
interactions. Journal
of Animal Ecology
55:923-940.
and J.Memmott.
2009.
Heleno, R. H., R. S Ceia, J.A. Ramos,
Effects of alien plants on insect abundance
and biomass: a
food-web approach. Conservation
Biology 23:410-419.
2008.
H., D. M. Bennett, and M. W. Cadotte.
a review of evenness effects on
of dominance:
Consequences
local and regional ecosystem processes.
89:1510
Ecology
1520.
R. D.
1977. Predation,
and
Holt,
apparent
competition,
structure of prey communities.
Theoretical
Population
Hillebrand,
12:197-229.
Biology
P. J., A. P. Dobson,
and K. D. Lafferty. 2006. Is a
Hudson,
in
healthy ecosystem one that is rich in parasites? Trends
21:381-385.
Ecology and Evolution
the impact of
Hulme, P. E., and E. T. Bremner. 2006. Assessing
on riparian habitats:
Impatiens
glandulifera
partitioning
diversity components
following
Ecology 43:43-50.
Applied

species
removal.
Journal of
1074
LUISA
R. M., and M. J. Crawley. 2002. Exotic plant invasions

the enemy release hypothesis. Trends
in Ecology
and
Keane,
and
Evolution
17:164-170.
Leibold, M. A., J.M. Chase,

turnover
1997. Species
structure. Annual
Review
467-494.
Liley,
G. CARVALHEIRO
D.,
and
development
nightjar
R.
and
T.
J. B. Shurin, and A. L. Downing.

the regulation
of trophic
of Ecology
and Systematics 28:
and
Clarke.
2003. The
impact of urban
on the numbers of
disturbance
on heathlands
in Dorset,
europaeus
human
Caprimulgus
114:219-230.
England. Biological Conservation
R. H.
1955. Fluctuations
of animal populations
MacArthur,
and a measure
of community stability. Ecology
36:533-536.
A. S., and R. Turkington.
2005. Are
invasive
MacDougall,
in degraded
of change
species the drivers or passengers
ecosystems? Ecology 86:42-55.
K. S. 2000.
McCann,
405:228-233.
The
diversity-stability
debate.
Nature
and M. V. Price. 2004. Tolerance

Memmott,
J., N. M. Waser,
of
of pollination networks to species extinctions. Proceedings
the Royal Society B 271:2605-2611.
T. E., J. C. Simpson, and M. D. Bertness. 2006.
Minchinton,
Mechanisms
of exclusion of native coastal marsh plants by
an invasive grass. Journal of Ecology 94:342-354.
A., and B. L. Li. 2008. Abundance
patterns inmulti
species communities exposed to habitat destruction. Journal
of Theoretical
Biology 251:593-605.
Morozov,
and H. C. J. Godfray.
2004.
R. J., O. T. Lewis,
evidence for apparent competition in a tropical
Experimental
forest food web. Nature 428:310-313.
R. Belshaw,
and H. C. J.
Muller, C. B., I. C. T. Adriaanse,
structure of an aphid-parasitoid
1999. The
Godfray.
Morris,
community. Journal of Animal Ecology 68:346-370.

Parker, J.D., D. E. Burkepile, and M. E. Hay. 2006. Opposing
effects of native and exotic herbivores on plant invasions.
Science 311:1459-1461.
Plantlife. 2007. Top 20 alien invasive species which threaten
the UK's
flora,
(http://www.plantlife.org.uk/uk/assets/
campaigning-for-change/invasive-plants/Top-20-most
threatening-in vasive-species.
pdf)
ET AL.
Ecology,
Vol.
91, No.
Price, P. W.,
and M. D. Hunter. 2005. Long-term population

dynamics of a sawfly show strong bottom-up effects. Journal
of Animal Ecology 74:917-925.
a language
R Development
Team.
Core
2007. R:
and
environment
for statistical computing. R Foundation
for
Statistical Computing, Vienna, Austria.
2007. Contrasting
effects of
Rand, T. A., and T. Tscharntke.
natural habitat loss on generalist and specialist aphid natural
enemies. Oikos
116:1353-1362.
count data in
S. A. 2008. Dealing with overdispersed
applied ecology. Journal of Applied Ecology 45:218-227.
tests of effects of plant
E.
1998. Experimental
Siemann,
productivity and diversity on grassland arthropod diversity.
Richards,
79:2057-2070.
Ecology
L. Durbin,
and F. Lowenstein.
2007.
Stinson, K., S. Kaufman,
invasion on a forest understory
Impacts of garlic mustard
14:73-88.
community. Northeastern Naturalist
2001. Empirical
and B. Wilsey.
Stirling, G.,
relationships
diver
between species richness, evenness, and proportional
158:286-299.
sity. American Naturalist
J.M. 2008. Understanding
the web of life: the birds,
Tylianakis,
the bees, and sex with aliens. Plos Biology 6:e47. [doi: 10.
1371 /journal.pbio.0060047]
and O. T. Lewis.
2007.
J. M.,
T. Tscharntke,
Tylianakis,
alters the structure of tropical host
Habitat modification
food webs. Nature 445:202-205.

parasitoid
L. L. Loope, M. Rejmanek,
Vitousek, P. M., C. M. D'Antonio,
1997. Introduced
and R. Westbrooks.
species: a significant
of human-caused
global change. New Zealand
component
Journal of Ecology 21:1-16.

and J.
Waser, N. M., L. Chittka, M. V. Price, N. M. Williams,
in pollination
1996. Generalization
Ollerton.
systems, and
why itmatters. Ecology 4:1043-1060.
a natural history of Britain's
Webb, N. R. 1986. Heathlands:
lowland heaths. Collins, London, UK.
1980. An ecological
survey of
Webb, N. R., and L. E. Haskins.
in 1978.
in the Poole Basin, Dorset,
heathlands
England,
17:281-296.
Biological Conservation
between adult
C.
1975. Evolutionary
Wiklund,
relationship
oviposition preferences and larval host plant range in Papilio
18:185-197.
machaon L. Oecologia
APPENDIX
Plot
lists and
locations, species
Archives E091-074-A1).
food webs,
comparison
of consumer/resource
models,
and
tests for apparent
(Ecological

competition

Diet Breadths and Invasive Plant Impacts On Food Webs

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Diet Breadths and Invasive Plant Impacts On Food Webs

Hochgeladen von

Copyright:

Verfügbare Formate

Diet breadth influences how the impact of invasive plants is propagated through food webs

Author(s): Luisa G. Carvalheiro, Yvonne M. Buckley and Jane Memmott

This content downloaded from 129.24.144.237 on Mon, 15 Apr 2013 17:26:53 PM

Ecology, 91(4), 2010, pp. 1063-1074

influences how the impact of invasive plants is

breadth influencesthepropagation of such impacts.Measuring the impact of the alien plant at

along with a decrease

communities differently,having significantand strongnegative impacts on the abundance of

considering changes in trophicdiversitywithin levels is hence essential for a full evaluation of

alien plants; apparent competition; Avon Heath Country Park, Dorset,

biodiversity (Vitousek et al.

1997), with numerous

Plants provide the

(Waser et al. 1996). As generalist and specialist guilds

positive effect of invasion, with the presence of alien

2006, Stinson et al. 2007). Among the few studies that

highly desirable goal as it could guide the development

propagation through natural communities need to be

plants leading to an increase of flower visitor frequency

2000) and while there are indications that diet breadth

2007), its role on impact propagation remains unclear

tained by grazing that,despite their low plant diversity,

This content downloaded from 129.24.144.237 on Mon, 15 Apr 2013 17:26:53 PM

species (Webb and Haskins 1980, Liley and Clarke

ceae) and Rhododendronponticum L. (Ericaceae) (Webb

North American heathland. This species is one of the

sought to identifyand quantify

were threefold. (1)We

the native community. Specialist herbivores from

introduced species that have become invaders (enemy

food web. As species at higher trophic levels are thought

communities. (3)We sought to testwhether diet breadth

ers are, themore likelytheyare to include species in their

1977). Therefore, we predict that diet

being less affected than specialist consumers, and that

specialistswill benefit from a release of natural enemies,

paths. The paths were mostly

bance changes (seeMacDougall

ulative experiments that add invasive species in a

proach utilizing a gradient of invasion, such as the one

is that it allows system stabilization. As most experi

resource/consumergenerations (Borer et al. 2005), they

extinction,and may be inadequate for studyingcompo

from 27 June to 15 September 2005. In each survey,one

all larval leaf herbivores were collected by beating the

All larvae were individually reared until an adult

pooled (rearingdata) or averaged (plant abundance) for

were either identifiedto the species levelormorphotyped

sufficientto distinguish parasitoids from hyperparasi

herbivore-parasitoid food webs for each plot, using

applied to control weeds, covering a gradient of G.

than are native plants?

Herbivory (measured as thenumber of individuals per

This content downloaded from 129.24.144.237 on Mon, 15 Apr 2013 17:26:53 PM

PROPAGATION OF IMPACTS THROUGH FOOD WEBS 1065

i.e., those present

in all eight studyplots: G. shallon,Calluna vulgaris (L.)

(herbivore density) were

log transformed and

in deviance during model

Objective 2: How does G. shallon abundance affect

invader, and a significant interaction between trophic

identifiedby combining our rearingdata with published