Oecologia (2007) 152:162–168
DOI 10.1007/s00442-006-0637-3
B E H A V IO R AL E C O L O GY
Phenotypically plastic responses of green frog embryos
to conXicting predation risk
D. H. Ireland · A. J. Wirsing · D. L. Murray
Received: 28 November 2006 / Accepted: 28 November 2006 / Published online: 10 January 2007
© Springer-Verlag 2007
Abstract Predators have been shown to alter the tim-
ing of switch points between life history stages, but few
studies have addressed switch point plasticity in prey
exposed simultaneously to conXicting predation pres-
sure. We tested hatching responses of green frog (Rana
clamitans) embryos subject to perceived predation risk
from chemical cues released by two stage-speciWc pre-
dators, predicting that these predators would elicit: (1)
directional hatching responses when presented inde-
pendently, and (2) intermediate phenotypic responses
when presented simultaneously. R. clamitans embryos
in outdoor exclosures were exposed to cues from an
egg predator (freshwater leeches; Nephelopsis obs-
cura), a larval predator (dragonXy nymphs, Aeschna
canadensis), and both predators in a 2 £ 2 factorial
experiment, and changes in hatchling size, hatchling
developmental stage, and hatching time were com-
pared to those for control embryos. Leeches alone
induced embryos to hatch at a smaller size and an ear-
lier developmental stage than controls, while dragonXy
nymphs elicited a delay in egg hatching time that was
associated with larger size and later developmental
Communicated by Steven Kohler.
D. H. Ireland · D. L. Murray
Department of Biology, Trent University,
1600 West Bank Drive, Peterborough, ON K9J 7B8, Canada
A. J. Wirsing (&)
Department of Biological Sciences,
Marine Biology Program, Florida International University,
Biscayne Bay Campus, AC1 388, 3000 NE 151 St,
North Miami, FL 33181, USA
e-mail: wirsinga@Wu.edu
123
stage at hatching. Embryos failed to respond to simul-
taneous exposure to both predators, implying that
responses to each occurred concurrently and were
therefore dampened. Our results indicate that prey
under threat from conXicting predators may manifest
intermediate defensive phenotypes. Such intermediate
responses may result in elevated rates of prey mortality
with possible consequences at the population level.
Keywords Chemical cues · Hatching traits ·
life history switch points · Phenotypic plasticity ·
Rana clamitans
Introduction
Studies of predator–prey interactions traditionally
have focused on pair-wise relationships (Relyea 2003),
but most prey species occur in environments where
multiple predators are present (Schoener 1989; Polis
1991). In such multi-predator systems, the eVects of
individual predator species on prey are rarely inter-
changeable (Polis and Strong 1996). Often, for exam-
ple, competition among predators and intra-guild
predation result in higher prey survival than that pre-
dicted by overall predator abundance (“risk reduc-
tion”; Sih et al. 1998). Alternatively, if phenotypically
plastic defensive responses by prey are tailored to spe-
ciWc predator species, then conXict between responses
to multiple predators may elevate prey vulnerability
(“risk enhancement”; Sih et al. 1998). A growing body
of literature suggests that these multiple-predator
eVects (MPEs) are commonly experienced by prey
within particular life history stages (see review by Sih
et al. 1998). However, the potential that MPEs span
Oecologia (2007) 152:162–168
diVerent life stages via induced shifts at switch points
has received less attention (see Vonesh and Osenberg
2003; Vonesh 2005; Vonesh and Warkentin 2006 for
examples).
Oviparous animals are often subject to diVerent
suites of predators as embryos and larvae (Vonesh
2005). Although they have few antipredator options,
embryos can escape egg predators by hatching early
(Warkentin 1995, 2000; Vonesh 2000; Chivers et al.
2001). However, prey individuals that hatch prema-
turely may be relatively small and, as a result, espe-
cially vulnerable to larval predators (Warkentin 1995).
It follows that predation may elicit plastic alterations in
the hatching characteristics of these species, with
hatching times and traits in particular habitats being set
in part by the intensity and direction of predator pres-
sure experienced by prey individuals in life history
stages adjacent to this switch point (Werner 1986;
Relyea 2003; Teplitsky et al. 2004, 2005). Predator-
induced plasticity in hatching attributes may have
important consequences for the dynamics of prey pop-
ulations (Sih and Moore 1993), so experiments assess-
ing hatching plasticity in the context of MPEs should
facilitate an improved understanding of both the struc-
ture of natural communities and the process of specia-
tion (Agrawal 2001). Although a variety of studies
have focused on responses of amphibian embryos to
perceived predation risk (e.g., Warkentin 1995, 1999;
Johnson et al. 2003; Vonesh and Osenberg 2003;
Vonesh and Bolker 2005), none have examined hatch-
ing plasticity by simultaneously exposing embryos to
multiple predators exerting conXicting pressure.
We explored phenotypic responses of green frog
(Rana clamitans) embryos to perceived risk from two
sequential, stage-speciWc predators—freshwater
leeches (Nephelopsis obscura) and dragonXy nymphs
(Aeshna canadensis). Both predator species frequent
R. clamitans oviposition sites. However, leeches com-
monly prey upon amphibian eggs, including those of
ranid frogs (Alford 1999; Chivers et al. 2001; Laurila
et al. 2002; personal observation). Thus, relative to
controls, we predicted that R. clamitans embryos would
respond to risk (chemical cues) from this predator with
accelerated hatching times and, as a result, reduced
size and developmental stage at hatching (Warkentin
1995; Vonesh 2000). In contrast, dragonXy nymphs
prey upon hatchling and tadpole prey (Skelley and
Werner 1990; Altwegg 2003; personal observation),
and are not known to consume eggs. Hence, we pre-
dicted that embryos exposed to risk from nymphs
would manifest delayed hatching times relative to con-
trols and, consequently, hatch at an advanced develop-
mental stage (Sih and Moore 1993). Since, the two
163
predators should favor conXicting phenotypic shifts, we
predicted that exposure to perceived risk from both
would result in: (1) an intermediate phenotype if each
elicited an independent and opposite response by
embryos, or (2) a directional phenotype if embryos can
discriminate between predators and optimize their
response by focusing on a single predator.
Materials and methods
Field collection
R. clamitans breeds in semi-permanent or permanent
ponds and lakes, with clutches containing 3,000–
3,500 eggs that are distributed as a surface Wlm in shal-
low waters at the riparian edge. R. clamitans eggs have
a short hatching time (3–5 days) compared to most
amphibian species (4–8 days after fertilization, see
Desrochers and Rodrigue 2004), and tadpole develop-
ment from hatching to metamorphosis usually requires
12–15 months (Duellman and Trueb 1986).
Over a 24-h period during July 2004, freshly depos-
ited R. clamitans egg masses were collected from three
semi-permanent ponds in southern Ontario (44°N,
78°W). Each pond was associated with a larger wetland
complex, and because ponds were separated by
<750 m, we assumed that resident green frogs in our
study area constituted a metapopulation. Pond temper-
ature reached »19°C near midday, and all ponds
contained N. obscura and high larval densities of dragon-
Xies from the genus Aeshna. Leeches were purchased
from local bait shops, while dragonXy nymphs were
collected from ditches near the collection ponds.
Experimental design
Three egg masses [<16 h old and at an earlier stage
than late cleavage (stage 9; Gosner 1960)] were trans-
ported to the laboratory and carefully split into 13 or
14 clumps of 100 eggs and placed in 19 l opaque plastic
tubs (40 £ 30 £ 19 cm3) Wlled with 16 l of Wltered river
water. Tubs were arrayed in a caged outdoor facility
and allowed to sit for 5 days prior to the addition of
eggs. Tubs were covered with window screen (1.5-mm
mesh size) for protection and shading. During the
experiment temperature ranged from 13° to 21°C. In
each tub, two cylindrical opaque plastic sections of
PVC pipe (15 £ 8 cm) were covered with a mesh win-
dow screen and placed in opposite corners of each tub.
Pipe sections served as cages for predators, thereby
eliminating any inXuence of direct predator eVects
(i.e., consumption) on eggs/embryos.
123
164
Our experiment employed a 2 £ 2 randomized
block design to allow for assessment of additive eVects
of predators; four treatments (Leech, Nymph,
Leech £ Nymph, Control) were replicated 10 times,
and three to four egg clumps from each clutch were
assigned to each treatment. Note that although they
are well established in the literature, additive designs
such as ours have been criticized because, in the multi-
ple predator treatment, predator density is confounded
with predator identity (Relyea 2003). However, we
elected not to use a substitutive approach entailing sin-
gle predator treatments with double the density as well
(Relyea 2003) because cue concentrations from two
predators of the same species are unlikely to resemble
those produced by two radically diVerent predator spe-
cies (i.e., in our view, single predator treatments at
double the density are not comparable to two predator
treatments). Thus, our experiment was designed spe-
ciWcally to assess embryo sensitivity to the presence,
rather than the magnitude, of the chemical cue associ-
ated with a particular predator species. The Leech
treatment included a single N. obscura fed ten R. clam-
itans embryos (stages 9–19; Gosner 1960), while the
Nymph treatment contained a late-instar A. canadensis
fed six R. clamitans hatchlings (stages 21–24; Gosner
1960); feeding occurred in the cage during a 48-h inter-
val prior to the onset of the experiment (i.e., day 0) to
ensure that any diet-borne chemical cues had been
excreted (i.e., were available for detection by prey dur-
ing the trials). Importantly, food rations given the two
predator species were of comparable biomass, mini-
mizing the likelihood that the dietary cues they pro-
duced diVered in concentration. The Nymph £ Leech
treatment contained both predators fed according to
the above rations, whereas the Control contained two
empty cages only. Predator cages were not disturbed
after the onset of the experiment, and no extra food
was provided [we recorded high survival for both leech
(100%) and nymph (95%) predators during the experi-
ment]. Predators were placed in the tubs 8 h after eggs
had been added.
On day 3, we began checking tubs every 8 h (0800,
1600, 2400 h) for recently hatched frog tadpoles.
Hatchling R. clamitans leave the vitelline membrane
and remain immobile on the substrate surface (Duell-
man and Trueb 1986), and we considered an embryo to
have hatched if it was found outside the membrane and
lying away from the egg mass. Time to hatching for a
given tub was calculated as the average time taken to
hatch by the Wrst ten individuals to meet the above cri-
teria. This sample is comparable to that used previ-
ously (Sih and Moore 1993; Schalk et al. 2002), and was
appropriate in this study because: (1) the experimental
123
Oecologia (2007) 152:162–168
unit was the tub and not the individual, (2) any bias
due to the method was consistent between treatments,
and (3) our a priori predictions were valid irrespective
of the speciWc sampling scheme. The ten hatched indi-
viduals per tub were preserved in 10% formalin; size
and developmental stage at hatching of preserved sam-
ples were assessed <5 days after the end of the experi-
ment. Total body length (tip of nose to tip of tail) was
measured to the nearest 0.25 mm using a dissecting
scope. Hatchling developmental stage was determined
according to Gosner (1960).
Statistical analyses
For each tub, we calculated the average hatching time,
hatchling size, and developmental stage at hatching.
Tadpole size, stage at hatching, and time to hatching
were normally distributed and thus untransformed
means were analyzed; developmental stage was
invoked as a continuous variable (Moore et al. 1996;
Chivers et al. 2001; Laurila et al. 2002). We employed a
multivariate ANOVA to evaluate overall treatment
eVects, with clutch origin serving as a block (three to
four replicates per clutch within each treatment; Schalk
et al. 2002). A two-way univariate ANOVA was then
used to elucidate the importance of all response vari-
ables that were at least marginally signiWcant
(P · 0.10) separately. The two-way ANOVA is the
most robust test for assessing interactive eVects
between treatments (Zar 1999), and we used
Leech £ Nymph interaction terms to determine if pre-
dators aVected embryo phenotype independently. We
considered signiWcant main eVects of predators, com-
bined with non-signiWcant interaction terms when both
predators were present, to constitute support for the
existence of independent predator responses. Means
for response variables in the Leech £ Nymph treat-
ment that were intermediate to those for main preda-
tor eVects (and comparable to means for the Control
treatment) were interpreted as support for our hypoth-
esis that the two predators used in the experiment
favor opposing hatching characteristics in R. clamitans.
Results
Multivariate analysis of variance revealed that eVect of
the Leech treatment on tadpole development was mar-
ginally signiWcant when considered in a multivariate
context (Table 1). Univariate ANOVA indicated that,
relative to controls, tadpoles hatched at a smaller size
(diVerence in means = 0.48 mm), at an earlier develop-
mental stage (diVerence in means = 0.32), and more
Oecologia (2007) 152:162–168
Table 1 Multivariate ANOVA (MANOVA) of hatching charac-
teristics and univariate ANOVA of hatching timing (Time),
hatching size (Size), and developmental stage at hatching (Stage)
Source MANOVA
Wilks’
F3,32 P Time
Leech 0.815 2.413 0.084
Nymph 0.665 5.36 0.004
Leech £ Nymph 0.952 0.538 0.66
Clutcha 0.737 1.756 0.122
a
Clutch was used as a block
swiftly (diVerence in means = 10.95 h) when eggs were
subject to the Leech treatment (Table 1; Fig. 1). In
contrast, the Nymph treatment increased tadpole size
(diVerence in means = 0.34 mm), advanced develop-
mental stage at hatching (diVerence in means = 0.29),
and lengthened egg hatching time (diVerence in
means = 10.95 h) (Table 1; Fig. 1).
Simultaneous exposure of R. clamitans eggs to the
Leech £ Nymph treatment did not aVect hatching
characteristics of tadpoles (Table 1), with sample
means for the two-species treatment being intermedi-
ate to those of treatments with each predator species
alone (Fig. 1). In general, means for the Leech £
Nymph treatment were comparable to those for the
Control group (diVerence in means between treated
and control eggs for tadpole size, developmental stage
at hatching, and hatching time were 0.16 mm, 0.14, and
6.39 h, respectively), implying that embryos responded
to both predators simultaneously.
Clutch origin had no eVect on hatching traits when
analyzed in a multivariate context (Table 1). However,
when assessed in a univariate context, tadpole size was
inXuenced signiWcantly by clutch (Table 1). Upon
hatching, larvae from the three clutches averaged
5.85 § 0.21, 5.85 § 0.13, and 6.35 § 0.13 mm in length,
respectively.
Discussion
The results of this investigation demonstrate that R.
clamitans embryos exhibit predator-speciWc phenotypic
responses, and that simultaneous exposure to multiple
predators imposing conXicting pressure elicits an inter-
mediate response. As predicted, embryos exposed to
perceived risk from egg predators (Leech treatment)
hatched at a small size and early developmental stage
relative to controls, while embryos exposed to risk
from larval predators (Nymph treatment) hatched
165
in green frog (Rana clamitans) embryos subjected to predation
risk (chemical cues) from freshwater leeches, dragonXy nymphs,
or both predators in a 2 £ 2 factorial experiment
ANOVA
Size Stage
F1,35 P F1,35 P F1,35 P
4.34 0.04 4.92 0.03 4.61 0.04
12.61 0.001 6.78 0.01 9.08 0.01
0.74 0.4 1.38 0.25 0.71 0.41
0.46 0.64 4.26 0.02 1.06 0.36
later, at a larger size, and at a more advanced develop-
mental stage than controls. However, embryos faced
with conXicting risks from egg- and larval-stage preda-
tors (i.e., those subjected to the Leech £ Nymph treat-
ment) failed to show an induced phenotypic shift,
instead manifesting hatching characteristics similar to
those of control embryos.
Eggs are an important prey item for many predators
(Chivers et al. 2001). Yet, because hatching has only
recently been identiWed as a potentially mutable anti-
predator defense, relatively few studies have addressed
the relationship between hatching traits and predation
(Chivers et al. 2001; Warkentin 2000). Among those
that have been conducted, our experiment is notable in
two ways. One, it constitutes a rare demonstration of
both advanced and delayed hatching by a single prey
species exposed separately to egg- and larval-stage pre-
dators. Further work is required to determine the
scope of such Xexibility. Two, it is the Wrst to reveal an
intermediate hatching response elicited by simulta-
neous exposure to conXicting, stage-speciWc predators.
In theory, organisms driven to manifest intermediate,
predator-speciWc defensive phenotypes by two preda-
tor species imposing conXicting pressure are predicted
to suVer from increased overall mortality (risk
enhancement; Lima 1992; Matsuda et al. 1993; Sih
et al. 1998). Thus, embryos and hatchlings in the R.
clamitans population we sampled may experience
higher rates of predation than those in populations that
are free from, or subject to a lesser degree of, conXict-
ing stage-speciWc predation pressure. If the extent to
which R. clamitans embryos hatching with intermedi-
ate characteristics are disadvantaged is dramatic
enough, in turn, then spatial heterogeneity in the abun-
dance of their predators may promote inter-population
divergence (Agrawal 2001), and inXuence the structure
of aquatic communities incorporating this amphibian.
Can the hatching responses to each individual pred-
ator observed in this experiment be used to estimate
123
166 Oecologia (2007) 152:162–168

7.0 a are adjusted to account for removal by predators, the

speciWc consequences of the shifts in hatching traits we
report (relative to controls) are diYcult to predict
6.5 based on the results of other studies involving anurans,
Hatchling siz e (mm)

especially since we exposed eggs only to predator cues.

+ Nymph
However, in general, variation in the size of recently
6.0 hatched anurans is known to dramatically aVect subse-
quent survival (Warkentin 1995). For example, Vonesh
- Nymph
and Bolker (2005) found that a 30% reduction in the
5.5 size of East African reed frog (Hyperolius spinigularis)
hatchlings altered larval mortality rates by a huge mar-
gin (90%), and that, after hatching, small (i.e., <10%)
5.0 increments in the size of recently emerged larvae cor-
responded with marked changes in survival. Therefore,
20.0 b
it is reasonable to assume that the 9% changes in
Developmental stage at hatching

hatchling size elicited by both the egg and larval preda-

tors used in this study are likely to be accompanied by
19.5
biologically meaningful changes in larval mortality
+ Nymph risk.
Chemical cues have been shown to elicit antipreda-
19.0
tor responses in a wide variety of taxa, including
- Nymph
amphibians (reviewed by Chivers and Smith 1998; Kats
18.5 and Dill 1998). However, this study is the Wrst to dem-
onstrate that, in the absence of consumption, conXict-
ing cues from multiple predators can combine to elicit
18.0 intermediate hatching characteristics. By implication,
the mere presence of multiple potential predators in
150 c particular areas (i.e., the threat of predation) may be
suYcient to induce hatching responses with conse-
quences for prey populations and, ultimately, commu-
Hatching time (hrs)

140
130
120
110
Leech absent
Fig. 1 a Hatching size, b developmental stage at hatching, and
c hatching time for Rana clamitans embryos with freshwater
leeches (Nephelopsis obscura) absent and present and with (open
circle, +Nymph) and without (Wlled circle, ¡Nymph) exposure to
dragonXy nymphs (Aeshna canadensis), means § SE
the degree to which R. clamitans embryos manifesting
intermediate phenotypes are disadvantaged? The
implications of predator-induced hatching plasticity for
prey survival are highly species-speciWc, and depend
heavily on the concomitant numerical eVects of the
predator in question (Vonesh and Bolker 2005). Thus,
in the absence of an empirical test where egg densities
+ Nymph nities. Given the speciWcity with which R. clamitans
embryos responded when exposed to each predator
separately, we dismiss the possibility that these chemi-
cal cues were derived solely from conspeciWc prey fed
to the predators prior to the onset of experimental tri-
- Nymph
als. Instead, we contend that the embryos responded to
cues emitted by the predators themselves (Kats and
Dill 1998). We acknowledge, however, that the inten-
sity of the observed phenotypic responses by R. clami-
Leech present
tans embryos to predation risk may have been predator
diet-dependent (Laurila et al. 1998; Chivers and Mirza
2001), and therefore that leeches and nymphs fed het-
erospeciWc prey may have elicited weaker shifts in
hatching traits. Accordingly, future work in this system
should seek to compare the magnitude of directional
hatching responses by R. clamitans to cues from preda-
tors fed conspeciWcs and heterospeciWcs (i.e., neutral
diets).
Previous studies have revealed marked inter- and
intra-population variability in predator-induced hatch-
ing responses (e.g., Schalk et al. 2002; Laurila et al.
2002). In general, the evidence presented here is not
consistent with this trend, as the measured hatching

123
Oecologia (2007) 152:162–168
traits did not diVer among broods (i.e., across space)
when evaluated in a multivariate context (Table 1).
However, we did Wnd that, in a univariate context, a
signiWcant amount of the variation in observed hatch-
ling size could be explained by brood origin (Table 1).
This result suggests that, for at least some traits, both
the selective regimes imposed by predators and the
costs associated with hatching plasticity experienced by
prey embryos may vary across Wne spatial scales. Not
surprisingly, such environmental heterogeneity is
viewed as a prerequisite for the evolution of plastic
antipredator traits (Clark and Harvell 1992; Tolrian
and Harvell 1999).
Responses by prey exposed simultaneously to multi-
ple predators are often predicted poorly by the sum of
pair-wise predator–prey interactions (e.g., Sih et al.
1998; Relyea 2003; Aukema et al. 2004; Teplitsky et al.
2004), implying that the results of experiments utilizing
single-predator environments may not always be appli-
cable to natural situations. Here, however, we demon-
strate that the independent and opposite eVects of two
predators on prey phenotype can dampen one another:
the directional outcomes of pair-wise interactions
between R. clamitans embryos and two stage-speciWc
predators were not observed when embryos were
exposed to cues from both predators simultaneously.
Thus, our study system provides an ideal opportunity
to test four additional, and largely unexplored, ques-
tions regarding predator-induced plasticity at life stage
interfaces:
1. Can an additive phenotypic response to multiple
predators become directional if relative risk is
altered on either side of the switch point (e.g., by
altering predator numbers or by presenting preda-
tors that vary in capture eYciency or motivation;
VanBuskirk 2002; Teplitsky et al. 2005)?
2. Can directional phenotypic responses be reversed
by removing predator cues at select stages during
embryo development or by introducing conXicting
predators sequentially?
3. At what point does increasing predation risk on
one side of a switch point cease to elicit a direc-
tional phenotypic response (i.e., when do the costs
associated with phenotypic alteration become pro-
hibitive)?
4. Does an intermediate response at a plastic switch
point necessarily translate into elevated mortality
(i.e., risk enhancement; Sih et al. 1998) relative to
defensive phenotypes that are not induced by the
presence of multiple predators?
Answers to these questions are central to our under-
standing of the means by which predators promote
167
individual phenotypic variation, changes in prey popu-
lation dynamics, and, ultimately, speciation.
Acknowledgements This study was funded by the National Sci-
ence and Engineering Research Council (NSERC) and the Can-
ada Research Chair (CRC) program. We are grateful to Kaitlin
Byrick for assistance with this research. This study was approved
by the Trent University Animal Care and Use Committee (proto-
col no. 04022). Steven Kohler and four anonymous reviewers pro-
vided useful comments on previous drafts of this manuscript.
References
Agrawal AA (2001) Phenotypic plasticity in the interactions and
evolution of species. Science 294:321–326
Alford RA (1999) Tadpole ecology: resource use, competition,
and predation. In: McDiarmid RW, Altig A (eds) Tadpoles:
the biology of anuran larvae. University of Chicago Press,
Chicago, Ill., pp 240–278
Altwegg R (2003) Hungry predators render predator avoidance
behavior in tadpoles ineVective. Oikos 100:311–316
Aukema BH, Clayton MK, RaVa KF (2004) Density-dependent
eVects of multiple predators sharing a common prey in an
endophytic habitat. Oecologia 139:418–426
Chivers DP, Mirza RS (2001) Importance of predator diet cues in
responses of larval wood frogs to Wsh and invertebrate preda-
tors. J Chem Ecol 27:45–51
Chivers DP, Smith RJF (1998) Chemical alarm signaling in aquat-
ic predator-prey systems: a review and prospectus. Eco-
science 5:338–352
Chivers DP, Kiesecker JM, Marco A, DeVito J, Anderson MT,
Blaustein AR (2001) Predator-induced life-history changes in
amphibians: egg predation induces hatching. Oikos 92:135–142
Clark CW, Harvell CD (1992) Inducible defenses and the alloca-
tion of resources: a minimal model. Am Nat 139:521–539
Desrochers JF, Rodrigue D (2004) Amphibiens et reptiles du
Quebec et des maritimes. Quintin, Waterloo, QC
Duellman WE, Trueb L (1986) Biology of amphibians. Mcgraw-
Hill, New York
Gosner KL (1960) A simpliWed table for staging anuran embryos
and larvae with notes on identiWcation. Herpetologica
16:183–190
Johnson JB, Saenz D, Adams CK, Connor RN (2003) The inXu-
ence of predator threat on the timing of a life-history switch
point: predator-induced hatching in the southern leopard
frog (Rana sphenocephala). Can J Zool 81:1608–1613
Kats LB, Dill LM (1998) The scent of death: chemosensory
assessment of predation risk by prey animals. Ecoscience
5:361–394
Laurila A, Kujasalo J, Ranta J (1998) Predator induced changes
in life history in two anurans: eVects of predator diet. Oikos
83:307–317
Laurila A, Pakkasmaa S, Crochet PA, Merilä J (2002) Predator-
induced plasticity in early life history and morphology in two
anuran amphibians. Oecologia 132:524–530
Lima SL (1992) Life in a multi-predator environment: some
considerations for antipredatory vigilance. Ann Zool Fenn
29:217–226
Matsuda H, Abrams PA, Hori M (1993) The eVect of adaptive
anti-predator behavior on exploitative competition and
mutualism between predators. Oikos 68:549–559
Moore RD, Newton B, Sih A (1996) Delayed hatching as a re-
sponse of streamside salamander eggs to chemical cues from
predatory sunWsh. Oikos 77:331–335
123
168
Polis GA (1991) Complex trophic interactions in deserts: an
empirical critique of food-web theory. Am Nat 138:123–155
Polis GA, Strong DR (1996) Food web complexity and commu-
nity dynamics. Am Nat 147:813–846
Relyea RA (2003) How prey respond to combined predators: a
review and an empirical test. Ecology 84:1827–1839
Schalk G, Forbes MR, Weatherhead PJ (2002) Developmental
plasticity and growth rates growth rates of green frog (Rana
clamitans) embryos and tadpoles in relation to a leech (Mac-
robdella decora) predator. Copeia 2002:445–449
Schoener TW (1989) Food webs from the small to the large. Ecol-
ogy 70:1559–1589
Sih A, Moore RD (1993) Delayed hatching of salamander eggs in
response to enhanced larval predation risk. Am Nat 142:947–
960
Sih A, Englund G, Wooster D (1998) Emergent impacts of multi-
ple predators on prey. Trends Ecol Evol 13:350–355
Skelly DK, Werner EE (1990) Behavioral and life-historical
responses of larval American toads to an odonate predator.
Ecology 71:2313–2322
Teplitsky C, Plénet S, Joly P (2004) Hierarchical responses of
tadpoles to multiple predators. Ecology 85:2888–2894
Teplitsky C, Plénet S, Joly P (2005) Costs and limits of dosage
response to predation risk to what extent can tadpoles invest
in anti-predator morphology? Oecologia 145:364–370
Tollrian R, Harvell CD (1999) The evolution of inducible defens-
es: current ideas. In: Tollrian R, Harvell CD (eds) The
ecology and evolution of inducible defences. Princeton
University Press, Princeton, N.J., pp 306–322
123
Oecologia (2007) 152:162–168
VanBuskirk J (2002) Dosage response of an induced defense: how
sensitive are tadpoles to predation risk? Ecology 65:1155–1160
Vonesh JR (2000) Dipteran predation on the eggs of four Hypero-
lius frog species in western Uganda. Copeia 2000:560–566
Vonesh JR (2005) Egg predation and predator-induced hatching
plasticity in the African reed frog, Hyperolius spinigularis.
Oikos 110:241–252
Vonesh JR, Osenberg CW (2003) Multi-predator eVects across
life-history stages: non-additivity of egg-and larval-stage pre-
dation in an African treefrog. Ecol Lett 6:503–508
Vonesh JR, Bolker BJ (2005) Compensatory larval responses
shift trade-oVs associated with predator-induced hatching
plasticity. Ecology 86:1580–1591
Vonesh JR, Warkentin KM (2006) Opposite shifts in size at meta-
morphosis in response to larval and metamorph predators.
Ecology 87:556–562
Warkentin KM (1995) Adaptive plasticity in hatching age: a re-
sponse to predation risk tradeoVs. Proc Natl Acad Sci USA
92:3507–510
Warkentin KM (1999) The development of behavioral defenses:
a mechanistic analysis of vulnerability in red-eyed tree frog
hatchlings. Behav Ecol 10:251–262
Warkentin KM (2000) Wasp predation and wasp-induced hatch-
ing of red-eyed treefrog eggs. Anim Behav 60:503–510
Werner EE (1986) Amphibian metamorphosis: growth rate, pre-
dation risk, and the optimal size at transformation. Am Nat
128:319–341
Zar JH (1999) Biostatistical analysis. Prentice Hall, Upper Saddle
River, N.J.