Ecology, 91(1), 2010, pp. 242–251
? 2010 by the Ecological Society of America
Sublethal predators and their injured prey: linking aquatic predators
and severe limb abnormalities in amphibians
JAY BOWERMAN,1PIETER T. J. JOHNSON,2,4AND TRACY BOWERMAN3
1Sunriver Nature Center, Box 3533, Sunriver, Oregon 97707 USA
2Ecology and Evolutionary Biology, University of Colorado, Ramaley N122, Campus Box 334, Boulder, Colorado 80309 USA
3Ecology Center, Watershed Sciences, Utah State University, Logan, Utah 84322 USA
blood or tissue without killing the prey, sometimes causing ecologically significant levels of
injury. We investigated the importance of sublethal predator attacks in driving an emerging
issue of conservation importance: missing-limb deformities in amphibians. We combined long-
term field data and manipulative experiments to evaluate the role of sublethal predation in
causing abnormalities in two regions of central Oregon, USA. Since 1988, western toads (Bufo
boreas) in Lake Aspen have exhibited abnormalities dominated by partially missing limbs and
digits at annual frequencies from ,1% to 35%. On Broken Top volcano, we found comparable
types and frequencies of abnormalities in Cascades frogs (Rana cascadae). Field sampling and
observational data implicated two aquatic predators in these abnormality phenomena:
introduced sticklebacks (Gasterosteus aculeatus) at Lake Aspen and corduliid dragonfly larvae
(Somatochlora albicincta) at Broken Top. In experiments, these predators produced limb
abnormalities identical to those observed in the respective regions. At Lake Aspen, in situ
predator exclosures effectively eliminated abnormalities in toads, while comparisons among
years with low and high stickleback abundance and between wetlands with and without
sticklebacks reinforced the link between fish and amphibian abnormalities. Neither trematode
parasite infection nor pesticide contamination could explain observed abnormalities. Our
results suggest that predators are an important explanation for missing-limb abnormalities
and highlight the ecological significance of sublethal predation in nature.
While many predators completely consume their prey, others feed only on
community ecology; Gasterosteus aculeatus; generalized estimating equations; invasive predators; limb
development; Oregon, USA; Rana cascadae; Somatochlora albicincta.
amphibian decline; amphibian deformities; amphibian malformations; Bufo boreas;
Predation is a central governing force in ecology and
evolution, often with wide-ranging effects on prey
survival, growth, reproduction, behavior, and pheno-
typic development. While ecological studies generally
focus either on how predators directly affect prey
survival or indirectly affect prey growth and behavior,
predator attacks can also cause sublethal injuries (e.g.,
Kaliszewicz 2003). Sublethal predator attacks often
occur when predators are smaller or similar in size to
their prey, have a limited gape size, or have specialized
mouthparts for extracting tissue or fluids (e.g., Schoener
1979, Pape-Lindstrom et al. 1997, Meyer and Byers
2005). Sublethal predation is well documented among
herbivores and among blood-feeding insects (e.g.,
mosquitoes, deerflies, reduviid bugs). Some sublethal
predators, however, can inflict severe and long-lasting
injuries with costly consequences for prey fitness.
Examples include sea lampreys, vampire bats, cookie
cutter sharks, and fangblennies. Mouritsen and Poulin
(2003), for example, reported that fish can attack and
injure up to 34% of marine cockles, resulting in between
9% and 21% loss of foot tissue per individual.
A better understanding of sublethal predation may
provide insights into the etiology of emerging conserva-
tion issues. Since the 1990s, increasing reports of
amphibians with skeletal abnormalities, including de-
formed, missing, and extra limbs, have caused concern
(Souder 2000, Ankley et al. 2004). Proposed causes of
amphibian limb abnormalities include chemical pollu-
tion, ultraviolet B radiation, predation, and parasite
infection (see Ouellet 2000, Ankley et al. 2004; Johnson
et al., in press). Morphological patterns may hold
important clues to proximate causative factors. The
trematode parasite Ribeiroia ondatrae, for example, can
cause missing limbs, bony triangles, skin webbings, and
duplicated structures (Sessions and Ruth 1990, Sessions
et al. 1999, Johnson et al. 1999, 2001a), but is rarely
associated with accounts involving predominantly miss-
ing limbs (Skelly et al. 2007, Reeves et al. 2008).
An important challenge in the continued study of
amphibian abnormalities is to identify the proximate
Manuscript received 11 September 2008; revised 20 March
2009; accepted 16 April 2009. Corresponding Editor: S. P.
causes of abnormalities in cases in which trematode
infection can be ruled out (see Lannoo et al. 2003).
Previous investigations have discounted predator-in-
duced limb abnormalities on the grounds that (1) a
predator would likely consume an entire frog rather
than just a leg, (2) injuries ought to produce obvious scar
tissue at the wound site, and (3) a frog is unlikely to
survive the traumatic loss of an entire limb (Meteyer et
al. 2000, Lannoo 2008). Although these assumptions
may hold for large vertebrate predators and for
amphibians injured after metamorphosis, they may not
apply to larval amphibians injured by smaller predators
such as aquatic insect larvae, leeches, small fishes, and
crayfish, all of which may be more likely to attack an
exposed portion of an amphibian such as a tail or a limb.
Tail injury can be severe (up to 75% loss) and
widespread (.30% prevalence) among larval amphibi-
ans (Feder 1983, Lawler et al. 1999, Blair and Wassersug
2000). Moreover, because larval amphibians have some
capacity for regeneration, partial or complete regrowth
following injury early in limb development might mask
signs of trauma and create the appearance of a
developmental malformation (e.g., Fry 1966). In Ger-
many, investigators showed that leech attacks caused
high prevalences of missing limbs or parts of limbs
(ectromelia) in common toads (Bufo bufo) (Viertel and
Veith 1992, Bohl et al. 1996, Bohl 1997). The role of
predators in causing amphibian limb abnormalities has
received little attention in North America, however.
We combined long-term monitoring data, field
observations, and mechanistic experiments to evaluate
the role of vertebrate and invertebrate predators in
causing missing-limb abnormalities in native amphibi-
ans. We evaluated the hypothesis that sublethal preda-
tion by either stickleback fishes (Gasterosteus aculeatus)
or dragonfly larvae (Somatochlora albicincta) was
responsible for high frequencies (.20%) of limb
abnormalities observed in metamorphosing western
toads (Bufo boreas) and Cascades frogs (Rana cascadae)
in central Oregon, USA. We further compared the
modes of attack of each predator and the types of
abnormalities induced to show that morphological clues
can help identify the predator responsible. Our results
help to account for missing-limb abnormalities in
amphibians and have important implications for under-
standing the significance of sublethal predation in
Abnormalities in western toads
Study sites.—Lake Aspen (‘‘LA’’; 4385302600N,
12182404000W) is an excavated impoundment of ;2 ha
with a maximum depth of 2.5 m that has served as a
major toad breeding site for 30 years (J. Bowerman,
personal observation). Nonnative three-spined stickle-
backs (G. aculeatus) were discovered in the Deschutes
River in 1980 (Oregon Department of Fish and Wildlife,
unpublished data) and quickly achieved high population
densities in LA that persist except during rare winterkill
events. Rear Driving Range pond (‘‘RDR’’; ,0.1 ha,
,1.5 m depth) supports sticklebacks only during
unusually mild winters, whereas Lake Penhollow
(‘‘LP’’; 6 ha, 7 m depth), Paulina Junction (‘‘PJ’’; ,0.5
ha, ,1.5 m depth), and Fairway 11 (‘‘FW’’; ,0.5 ha,
,1.5 m depth) ponds are isolated from other water
bodies and remain stickleback-free. These ponds are
excavations within 15 km of LA that intermittently
support toad reproduction. Lake Penhollow did not
exist prior to 1999, PJ only intermittently produces
toads, and FW last supported toads in 1997.
Field observations.—Since the discovery of abnormal
toads at Lake Aspen in 1988, we have inspected a total
of 13443 metamorphic toads representing 11 seasons of
major toad emergences at this site. To avoid bias that
might result from greater mobility of normal vs.
abnormal toads, all sampling was conducted during
emergence events and within 10 m of the shoreline.
Metamorphic toads were either hand-captured or netted
from the water, held in buckets until inspection, then
released within 10 m of capture location. Subsequent
collections within the same year were conducted .100 m
from prior collections to avoid resampling the same
individuals. For comparison with Lake Aspen, we
inspected a minimum of 200 emerging toads from each
of the other four wetlands during opportunistic visits
when these ponds supported toad reproduction, using
the same inspection techniques (LP, RDR, and PJ in
2001 and FW in 1997). We visually inspected meta-
morphs by extending the hind limbs and comparing the
shape, size, and position of all limb parts. We classified
deformities into four categories of increasing severity of
limb truncation: missing toes, missing foot, partial leg
loss (between foot and femur), and whole leg loss (some
or all visible portions of femur missing).
We also sampled LA for several factors with the
potential to cause abnormalities: predators, trematodes,
and pollutants. We conducted coarse-level monitoring
of stickleback populations using Gee’s minnow traps
(Cuba Specialty Manufacturing, Fillmore, New York,
USA) at two locations within Lake Aspen. Traps were
deployed for 24 h intermittently between May and
August of each year. We classified sticklebacks as
‘‘abundant’’ (.100 individuals?trap?1?d?1) or ‘‘rare’’
(,10 individuals?trap?1?d?1). In a subset of years,
including two years following severe winterkill events,
we enumerated all captured fish to calculate a catch per
unit effort (CPUE) and validate our coarse-level
estimates. We necropsied a subset (n ¼ 5) of abnormal
toads to quantify parasite infection with a focus on
Ribeiroia abundance. We also tested water for 61
different pesticides and metabolite breakdown products
(e.g., herbicides and organophosphates) as part of
another study (Johnson et al. 2002).
Predator exclosures.—To evaluate whether isolating
toad larvae from aquatic predators produced differences
in tail and/or limb condition, we constructed two 1-m3
January 2010243PREDATION AND ABNORMAL AMPHIBIANS
exclosures (13230.5 m) made of woven nylon mesh (5-
mm openings) and suspended from floating log frames.
This effort was not intended as a rigorous, replicated
experiment, but rather as a qualitative comparison of
larval toad development inside and outside the exclo-
sures. Each exclosure received 500 field-caught tadpoles
(stage 26 [Gosner 1960]) with intact caudal fins. Twice
weekly we added aquatic plants (Elodea) in sufficient
quantity to create a density similar to the elodea beds
adjacent to the exclosures (after removing any stickle-
backs in the plant material). Mesh openings were large
enough to allow entrance of water-borne chemicals or
free-swimming trematode cercariae. After two weeks
(Gosner stage 30), we compared the tail condition of 400
toad larvae within the exclosures with 400 free-
swimming larvae captured within 10 m of the exclosures,
replacing all individuals after inspection. Garter snakes
(Thamnophis sirtalis) invaded the exclosures during week
6 and consumed some of the tadpoles, necessitating that
the experiment be terminated. We inspected the 120
remaining larvae (Gosner stages 38–39) from the
exclosures and compared their tail and hind-limb
condition with those of 400 free-swimming larvae
collected adjacent to the exclosures. Snake predation
was unlikely to have altered the nature of our results
because limb development was too early to affect escape
maneuvers and caudal fin reductions representing ,5%
of tail area (as seen within enclosures) typically have
negligible effects on either maximum speed or escape
time (Van Buskirk and McCollum 2000).
Aquarium experiment.—We investigated interactions
between sticklebacks and toad larvae in glass aquaria
(48 3 25 3 30 cm) containing 30 L of well water and a
small quantity of Elodea. Each tank received five B.
boreas larvae (Gosner stages 35–38) without visible
abnormalities. Predation treatments received four adult
sticklebacks captured from Lake Aspen and acclimated
in aquaria for 2–3 d prior to the experiment. Control
aquaria were treated identically but received no stickle-
backs. After 24 h, we removed and inspected all
tadpoles. We used fish and tadpoles only once and
rotated aquaria between experimental and control
treatments. We conducted eight replicates of each
treatment (16 trials in total) and observed experimental
and control aquaria for a total of 4 h during
experiments, noting fish and tadpole behavior. A subset
of tadpoles that lost limbs during experiments was raised
Abnormalities in Cascades frogs
Study sites.—Broken Top Ponds 1 and 2 (BT1 and
BT2) are small (,0.1 ha), shallow (,1 m), high-
elevation wetlands (2200 m) located in Deschutes
County, Oregon (4480300400N, 12184104000W). Water is
uniformly clear over a substrate of fine sediment and
airfall pumice with ,1% vegetation cover. Both ponds
support breeding Cascades frogs (Rana cascadae) and
long-toed salamanders (Ambystoma macrodactylum),
but lack fish or snails.
Field observations.—In September 2003, we found a
high frequency (27.5%) of R. cascadae larvae with limb
abnormalities consisting of missing or partially missing
hind limbs. We sampled BT1 and BT2 twice in 2003 and
twice in 2004, recording the prevalence and composition
of abnormalities as described for examining toads.
Using a D-frame dipnet (0.5-m opening30.5-cm mesh),
we performed 2-m substrate sweeps at 11 evenly spaced
points around each pond’s perimeter and two additional
sweeps 6 m offshore. Captured organisms were identi-
fied, counted, and released. Five abnormal R. cascadae
were necropsied to assess parasite infection.
Predator experiment.—We identified two predators
with the potential to cause limb loss in R. cascadae:
larval long-toed salamanders (Ambystoma macrodacty-
lum) and larval dragonflies (Somatochlora albicincta).
Naiads of S. albicincta consume aquatic insects,
freshwater shrimp, small fishes, and tadpoles (Corbet
1999). Long-toed salamanders are known to attack
other amphibians and sometimes cause limb damage
(see Wildy et al. 2000, Johnson et al. 2006). We
compared the effects of salamander larvae and dragonfly
naiads on R. cascadae limb development within covered
storage containers (65 3 35 3 15 cm), each of which
received 5 cm of pond water, 1 cm of mud (strained
through a 0.5-cm filter to remove any potential
predators), and 10 R. cascadae larvae (Gosner stages
35–40; 20.9 mm mean snout-vent length [SVL]).
Containers were assigned randomly to one of three
treatments with 21 replicates per treatment: dragonfly
larvae (n ¼ 5, 21.3 mm mean length), salamander larvae
(n¼5, 32.7 mm mean SVL), or no predators. All larvae
were captured and examined immediately prior to each
trial. We maintained containers in sheltered locations
near the ponds and rotated them among treatments and
positions between trials. Containers were rinsed thor-
oughly between trials and neither predators nor larval R.
cascadae were used more than once. After 48 h, we
removed and inspected all R. cascadae larvae for
abnormalities. A subset of abnormal R. cascadae larvae
was raised through metamorphosis while remaining
predators and tadpoles were released.
Analysis.—To compare the frequency of abnormali-
ties among toad larvae from inside and outside the
exclosures, we used a contingency table analysis that
evaluated whether the status of each animal (normal vs.
abnormal) was independent of its source (free-living vs.
protected). For experiments involving sticklebacks in
aquaria and salamander or dragonfly predators in
outdoor chambers, we used generalized estimating
equations (GEE). Generalized estimating equations are
an extension of the generalized linear model that allow
for correlations among observations from the same
subject (e.g., individuals within the same container) and
use an iterative approach to estimate the correlations
among observations (Liang and Zeger 1986, Horton and
JAY BOWERMAN ET AL. 244Ecology, Vol. 91, No. 1
Lipsitz 1999). Generalized estimating equations are most
often used for longitudinal or clustered data when the
outcome variable is discrete or binary. Because multiple
tadpoles were maintained in containers with predators,
the likelihood of one tadpole experiencing an injury may
be influenced by the fate of co-occurring tadpoles (i.e., if
predators become satiated after feeding on one tadpole).
We therefore used GEEs to explicitly allow for
correlations among individual tadpoles within a single
container and evaluated the effect of predator treatment
on the likelihood of limb abnormality (abnormal vs.
normal) or death (dead vs. alive) during the experimen-
tal trials. We used the procedure ‘‘geepack’’ (Halekoh et
al. 2001) in the statistical package R (R Core
Development Team, Vienna, Austria). We assumed an
unstructured covariance matrix for our data and
evaluated the significance of each treatment using
ANOVA following the methods outlined in Halekoh et
Abnormalities in western toads
Field patterns.—Between 1988 and 2008, the preva-
lence of limb deformities among toad metamorphs at
Lake Aspen ranged from ,1% to 34.2% (12.9 6 3.9
[mean 6 SE]; n¼13,443; Fig. 1A). We typically observed
tail damage in larval B. boreas by Gosner stage 30
whereas limb abnormalities became evident in later-stage
larvae and metamorphs. Tail damage consisted of small
tears or nicks to the margin and typically affected ,5% of
total fin area. In most years, sticklebacks were abundant
in LA (.100 sticklebacks?trap?1?d?1). In 1993 and 2008,
however, following severe winterkill events, we caught an
average of 1.7 and 8.5 sticklebacks?trap?1?d?1, respec-
tively, compared to 176 sticklebacks?trap?1?d?1in 1998
and 104 sticklebacks?trap?1?d?1in 2003. Corresponding-
ly, abnormality frequencies in 1993 and 2008 were 0.05%
(n ¼ 2000) and 1.2% (n ¼ 1016), respectively, which
contrasted sharply with the average abnormality fre-
quency of 15.6% in years with typical stickleback
abundances (Fig. 1A). Comparative data from surround-
ing water bodies in Deschutes County supported the link
between limb abnormalities in toads and the occurrence
of sticklebacks. In RDR, which supported an average of
90 sticklebacks?trap?1?d?1in 2001, 12.3% of examined
toads exhibited limb abnormalities, whereas ,1% of
toads were abnormal among wetlands without stickle-
backs (Fig. 1B).
Toad abnormalities from LA and RDR ranged from
missing digits to partially and completely missing limbs
(Fig. 2). Missing digits and feet were most common,
(65.1% of all abnormalities), followed by partially
(22.2%) and completely missing limbs (7.9%; Table 1).
Bilateral deformities were uncommon and were rarely
symmetrical when they occurred. Among more than
1000 abnormal toads observed, we saw only two
instances of polymely and no individuals with bony
triangles, skin webbings, or other abnormalities charac-
teristic of trematode-related malformations (see Johnson
et al. 1999, 2001a). Necropsies of five abnormal toads
from LA detected Ribeiroia in only one individual. This
observation, combined with the absence of more typical
parasite-related malformations (e.g., bony triangles and
skin webbings) suggests that trematode infection did not
contribute significantly to the observed pattern of
abnormalities (see detailed discussion in Johnson et al.
2001a). No pesticides or pesticide metabolites were
detected in the water analyses (Johnson et al. 2002).
Predator exclosures.—Predator exclosures in LA
eliminated limb and tail abnormalities in nearly all toad
larvae. Among unprotected larvae from LA, 12.3% of
early-stage individuals and 89% of late-stage larvae
exhibited ragged caudal fin margins, whereas 0.8% of
early-stage and 1.7% of late-stage toads from exclosures
exhibited any tail damage (Pearson v2¼ 331.3, df ¼ 1,
two-tailed P , 0.0001). Similarly, while 4.5% of free-
living larvae exhibited limb abnormalities, all larvae
inside the exclosures had normal limbs (Pearson v2¼
5.59, df ¼ 1, two-tailed P , 0.02; Fig. 3).
Aquarium experiment.—Among 40 tadpoles exposed
to sticklebacks in aquaria, 53% exhibited tail damage
and 7.5% suffered hind-limb abnormalities compared to
2.5% of individuals with tail damage and 0% with limb
abnormalities among tadpoles maintained only with
conspecifics. Tail and limb damage among larvae in the
tanks with sticklebacks was consistent with field
observations and yielded post-metamorphic deformities
indistinguishable from those seen in wild-caught indi-
viduals. Results of the GEE analysis revealed a strong
effect of sticklebacks on the risk of limb and tail damage
in larval toads (limbs, estimate of coefficient ¼ 42.82 6
0.61, Wald statistic ¼4890.7, P , 0.0001; tails, estimate
of coefficient ¼ 3.34 6 1.11, Wald statistic ¼ 9.15, P ,
Abnormalities in Cascades frogs
Field observations.—At Broken Top Ponds 1 and 2,
we observed a high prevalence (5–21%) of hind-limb
abnormalities among pre- and post-metamorphic Cas-
cades frogs (Fig. 4A). The types of abnormalities were
similar to those observed in LA toads but often more
severe, with limb abnormalities in Cascades frogs often
affecting .50% of a limb (e.g., missing tibiafibula and
portion of femur). We also observed several cases of
micromelia (limb reduced in size but otherwise com-
plete). Tail abnormalities occurred in ,20% of late-stage
tadpoles from BT1 and BT2.
Dipnet sweeps of substrate captured an average of 3.7
dragonfly larvae and 0.8 salamander larvae per sweep.
On several occasions, we observed dragonfly larvae
clasping the hind limb of individual R. cascadae
tadpoles. Although the much heavier tadpoles would
periodically thrash violently, none managed to dislodge
the dragonfly. Necropsies performed on five deformed
metamorphs from BT1 found no Ribeiroia parasites and
the absence of snails from BT1 or BT2 further ruled out
January 2010 245PREDATION AND ABNORMAL AMPHIBIANS
trematode infection as a cause of these abnormalities.
No water quality analyses were performed at these
Container experiments.—Of the 210 tadpoles main-
tained with dragonfly larvae, 37 (17.6%) suffered
complete or partial amputation of one or both hind
limbs (Fig. 4B). Four tadpoles lost both hind limbs. In
contrast, two tadpoles in containers with salamander
larvae and one from the control exhibited limb
abnormalities. The GEE analysis indicated a significant
effect only for dragonfly larvae, with no difference
between the salamander and control treatments (drag-
onfly treatment, coefficient estimate¼2.25 6 0.73, Wald
statistic ¼ 9.43, P , 0.005; salamander treatment,
coefficient estimate ¼ 0.75 6 0.87, Wald statistic ¼
0.74, P ¼ 0.3). Of the 37 abnormal hind limbs produced
abnormalities in metamorphosing toads from Lake Aspen (1988–2008). Only years with successful toad emergence are displayed.
Bars in black with vertical arrows above them indicate years in which the abundance of three-spined sticklebacks (Gasterosteus
aculeatus) was low (,10 individuals?trap?1?d?1) due to winterkill events. In all other years, stickleback abundance was estimated at
or above 100 individuals?trap?1?d?1. (B) Comparison of limb abnormality frequency in wetlands as a function of stickleback
abundance (CPUE is catch per unit effort). The wetland sites Lake Aspen (LA) and Rear Driving Range (RDR) pond supported
high densities of sticklebacks whereas Lake Penhollow (LP), Paulina Junction (PJ), and Fairway 11 (FW) lacked sticklebacks in the
year of sampling. Error bars in both panels represent the 95% confidence interval for a proportion.
Pattern of limb abnormalities in western toads (Bufo boreas) in central Oregon, USA. (A) Frequency of limb
JAY BOWERMAN ET AL.246 Ecology, Vol. 91, No. 1
in the dragonfly treatment, 22 involved amputations of
an entire hind limb, 13 amputations removed more than
50% of a hind limb, and two resulted in loss of a hind
foot. Tadpoles with amputations healed quickly leaving
no sign of scarring by metamorphosis.
After more than a decade of focused attention on
amphibian deformities, some causes of malformations,
such as the trematode parasite Ribeiroia (Johnson et al.
1999, 2002, 2006, Sessions et al. 1999, Kiesecker 2002,
Stopper et al. 2002) have been well-documented.
However, despite numerous reports of abnormalities
involving predominantly missing or truncated limbs in
the absence of Ribeiroia infection, few studies have
explored the contributing role of sublethal predation.
Although some researchers have suggested that preda-
tors are unlikely to explain limb abnormalities, citing the
improbability of a predator removing only the limb of a
frog and the absence of obvious scar tissue in abnormal
animals (e.g., Meteyer et al. 2000, Lannoo 2008), we
suggest that the hind limbs of larval anurans may be
boreas) with typical missing-limb abnormalities (scale in cm); (B) toad with a completely missing hind limb; (C) sticklebacks
(Gasterosteus aculeatus) from Lake Aspen; (D) jaws and teeth of a stickleback with overlying tissue removed.
Representative limb abnormalities and predators from Lake Aspen, Oregon: (A) metamorphic western toads (Bufo
(Bufo boreas) from Lake Aspen and in metamorphosing Cascades frogs (Rana cascadae) from
Broken Top ponds in central Oregon, USA.
Distribution of limb abnormalities found in a subset of metamorphosing western toads
Lake Aspen (Bufo boreas) Broken Top (Rana cascadae)
of all deformitiesNo. deformities
of all deformities
Partially missing limb
Total 343100 58100
? Defined here as more than two-thirds of limb missing.
January 2010 247PREDATION AND ABNORMAL AMPHIBIANS
particularly vulnerable to predator attack. Because the
hind limbs of larval anurans lack hard skeletal elements
and become increasingly exposed over time, they may be
especially susceptible to predators that are gape-limited
or too small to consume an entire larva. Moreover, the
partial regenerative ability of larval anurans may
obscure obvious signs of trauma such as scarring.
Results from long-term field observations and exper-
iments indicate that sublethal predation by vertebrate
and invertebrate predators can contribute to missing-
limb abnormalities in amphibians. Three lines of
evidence point to sublethal predation by sticklebacks
as the principle cause of the limb deformities in studied
toad populations. First, toad deformities were most
common at sites with abundant sticklebacks but rare at
nearby wetlands lacking sticklebacks. Moreover, follow-
ing severe winterkill events that reduced stickleback
abundances within Lake Aspen, toad deformities
declined to low levels (,5%) (Fig. 1). Second, prelim-
inary in situ predator exclosures effectively eliminated
tail damage and limb deformities among developing
toads, suggesting that neither water quality nor trema-
tode infection was involved (observations that were
supported by pesticide analyses and parasite examina-
tions). We acknowledge, however, that invasion by
garter snakes into the exclosures make such results
preliminary and in need of additional confirmation.
Finally, sticklebacks attacked tadpoles in aquaria and
produced damage to tails and hind limbs consistent
with, and indistinguishable from, abnormalities ob-
served in the field.
In two high-elevation wetlands on Broken Top
volcano, metamorphosing Cascades frogs exhibited high
frequencies (5–25%) of missing limbs, partially missing
limbs, and shrunken limbs (micromelia). In this case,
direct observations and predation experiments suggested
that the causative agent was a corduliid dragonfly larva
(Somatochlora albicincta). While larval dragonflies such
as Aeshna and Anax spp. are widely recognized as
important lethal predators of larval amphibians (e.g.,
Smith 1983, Wilbur and Fauth 1990), this is the first
report of which we are aware linking odonate predation
to mass limb abnormalities in amphibians. Unlike the
penetrating lapial palps of Aeshna and Anax spp.,
corduliid and libellulid dragonfly larvae have palps
better suited for clasping and cutting (Fig. 4D), possibly
contributing to their disproportionate effects on devel-
oping limbs. Further, larval Rana cascadae are large
relative to S. albicincta, which may explain why injury
was a more common outcome than death in experi-
ments. Neither trematode parasites nor other predators
could account for observed abnormalities.
Among abnormal toads and Cascades frogs, we rarely
observed obvious signs of trauma such as scarring
following metamorphosis. Anurans have varying de-
grees of regenerative ability that decline as larval
development proceeds (e.g., Kurabuchi and Inoue
1982), which may help to explain the lack of visible
trauma. Fry (1966) found that if leopard frog (R.
pipiens) limbs were amputated prior to differentiation of
the digits, injured limbs regenerated into fully patterned
albeit smaller limbs (micromelia); when amputation
occurred later in development, however, limbs healed
into stumps lacking joints, toes, or other limb-like
features. We observed all of these conditions in Cascades
frogs (Fig. 4). The absence of rigid, calcified skeletal
elements in developing tadpole limbs, combined with
their limited regenerative capacity, may help explain
why observers seldom see evidence of injury, such as
protruding bone or scarring. Nonetheless, while stickle-
backs and dragonfly larvae caused generally similar
deformities in anurans, deformities caused by stickle-
backs primarily affected the toes and feet of toads,
wetlands of Broken Top volcano, Oregon. (A) Percentage of
limb abnormalities in larval and metamorphic Cascades frogs
(Rana cascadae) inspected from Broken Top (BT) ponds 1 and
2 in both 2003 and 2004. Error bars represent the 95%
confidence interval for a percentage. (B) Results of experimen-
tal trials examining the effect of predator identity on limb
development in Cascades frog larvae. Illustrated is the
percentage of amphibian larvae with missing-limb abnormali-
ties as a function of predator treatment, including no predator
(control), exposure to larval long-toed salamanders (Ambysto-
ma macrodactylum), and exposure to larval dragonflies (Soma-
tochlora albicincta). Predation trials lasted 48 h. Error bars
Results of field and experimental studies from
JAY BOWERMAN ET AL.248 Ecology, Vol. 91, No. 1
whereas dragonfly attacks more often removed all or
most of the hind limb and only rarely caused the loss of
toes or feet (Table 1). This difference likely reflects
differences in the anatomy and attack behavior between
the respective predators: sticklebacks attack rapidly and
tear away small pieces of tissue whereas corduliid
dragonfly larvae capture and hold prey with their labial
palps before chewing through prey tissue with their
mandibles (see Figs. 2D and 4D).
We suggest that sublethal predation may be an
important explanation of missing-limb abnormalities in
other amphibian populations. Working in the United
Kingdom and the Netherlands, Wisniewiski (1958) and
Van Gelder and Strijbosch (1995), respectively, each
documented large numbers of common toads (Bufo
bufo) with missing-limb abnormalities. The specific
causes of the abnormalities were not isolated, but both
groups suspected predators might be important contrib-
utors. In Panama, Duellman and Trueb (1994) repeat-
edly observed crabs attack frogs (Colostethus inguinalis),
often resulting in missing limbs. Gray et al. (2002)
similarly suggested the involvement of crabs in explain-
ing high levels (12%) of digit abnormalities in the green
poison dart frog (Dendrobates auratus), also in Panama.
In an unpublished study, C. Corkran and colleagues
noted missing-limb abnormalities in .40% of metamor-
phosing toads from Frog Lake, Oregon. Based on the
composition of the abnormalities and field observations,
the researchers suggested that the introduced speckled
dace (Rhinichthyes osculus) was responsible for the
abnormalities (C. Corkran, personal communication).
Studies of leech (Erpobdella octoculata) attacks on
common toads (Bufo bufo) in Germany provide some of
the best evidence of the importance of sublethal
predation in causing amphibian abnormalities (Viertel
and Veith 1992, Veith and Viertel 1993, Bohl et al. 1996,
Bohl 1997). High frequencies (.20%) of partially and
completely missing limbs in metamorphic toads were not
associated with genetic anomalies or contaminants;
however, experimental exclosures permeable to water
but excluding predators eliminated the abnormalities
(Bohl 1997). Further, experimental exposure to leeches
in the laboratory induced the same suite of abnormal-
ities in toads (Viertel and Veith 1992). An experimental
reduction of leech density within a wetland sharply
reduced abnormality levels in the toad population,
offering ecosystem-level evidence for the causal role of
leeches (Bohl et al. 1996).
abnormalities in metamorphosing Cascades frogs (Rana cascadae), illustrating partially missing limbs, completely missing limbs,
and shrunken limbs (micromelia); (B) Rana cascadae larva with proximal portion of left leg clasped by labial palps of a larval
dragonfly (black arrow); (C) larval dragonfly (Somatochlora albicincta); (D) comparison of the labial palps of S. albicincta (left) and
an aeshnid dragonfly (right), contrasting the clasping apparatus of the corduliid with the piercing apparatus of the aeshnid. Scale
bars indicate 1 cm.
Representative abnormalities and predators from wetlands of Broken Top volcano, Oregon: (A) typical limb
January 2010249PREDATION AND ABNORMAL AMPHIBIANS
A pressing question surrounding the occurrence of
predation-related injuries is whether they pose a threat
to affected individuals and populations. Meyer and
Byers (2005), for example, reported the sublethal
predation injuries to the siphons of marine bivalves.
Shortened siphons forced bivalves to remain at shallow
depths in the sediment, thereby making them more
vulnerable to lethal attacks by excavating predators.
Similarly, severe tail injury in tadpoles can significantly
increase their vulnerability to turtle and crayfish
predators (Feder 1983, Figiel and Semlitsch 1991). We
suspect that observed limb abnormalities substantially
reduce survivorship among affected amphibians (see
Johnson et al. 2001b). Unlike tail injuries, which
disappear at metamorphosis, limb abnormalities are
most likely to reduce the survival of post-metamorphic
In summary, our results indicate that attacks by
vertebrate and invertebrate predators can cause high
frequencies of missing-limb abnormalities similar to
those observed in nature. Considering that amphibians
with missing limbs and digits are among the most
commonly reported anuran abnormalities in the United
States (e.g., Guderyahn 2006), we suggest that the role
of predators deserves further investigation. Although
there are other causes of missing limbs in amphibians
besides sublethal predation, the potential involvement of
predators, which can often be examined through
relatively inexpensive techniques (e.g., exclosure stud-
ies), should not be discounted a priori. Sublethal
predation and the resulting injuries in amphibians may
be important not only as ecological or evolutionary
influences, but may also have significant implications for
conservation and contemporary concerns over wide-
spread amphibian population losses (Stuart et al. 2004).
While predators are a ubiquitous component of most
aquatic ecosystems, not all amphibian populations
exhibit high frequencies of limb abnormalities. Thus,
determining which predators cause sublethal injury and
under what conditions they are most likely to do so are
important priorities in the continued study of amphibian
abnormalities. We emphasize the need for more work to
understand (1) under what circumstances sublethal
predation causes higher-than-baseline levels of abnor-
malities and (2) the consequences of sublethal predator
attacks for amphibian ecology and conservation.
We thank Margaret Beard, Michael Bell, Michael Blouin,
Chuck Kimmel, and Stan Sessions for comments and sugges-
tions. Thoughtful critique and suggestions by two anonymous
reviewers improved the manuscript.
Ankley, G. T., S. J. Degitz, S. A. Diamond, and J. E. Tietge.
2004. Assessment of environmental stressors potentially
responsible for malformations in North American anuran
amphibians. Ecotoxicology and Environmental Safety 58:7–
Blair, J., and R. J. Wassersug. 2000. Variation in the pattern of
predator-induced damage to tadpole tails. Copeia 2000:390–
Bohl, E. 1997. Limb deformities of amphibian larvae in Aufsess
(Upper Franconia): attempt to determine causes. Mu ¨ nich
Contributions to Wastewater Fishery and River Biology 50:
Bohl, E., E. Bohl, J. Heise, R. Fischer, and M. Popp. 1996.
Untersuchungen zu Mißbildungen an Amphibienlarven im
Natureich des Beispielsbetriebs in Aufseß (Oberfranken).
Report to the Bayerisches Landesamt fu ¨ r Wasserwirtschaft,
Institut fu ¨ r Wasserforschung, Mu ¨ nchen, Germany.
Corbet, P. S. 1999. Dragonflies: behavior and ecology of
Odonata. Cornell University Press, Ithaca, New York, USA.
Duellman, W. E., and L. Trueb. 1994. Biology of amphibians.
Johns Hopkins University Press, Baltimore, Maryland, USA.
Feder, M. E. 1983. The relation of air breathing and
locomotion to predation on tadpoles, Rana berlandieri, by
turtles. Physiological Zoology 56:522–531.
Figiel, C. R., and R. D. Semlitsch. 1991. Effects of nonlethal
injury and habitat complexity on predation in tadpole
populations. Canadian Journal of Zoology 69:830–834.
Fry, A. E. 1966. Hind limb regeneration in Rana pipiens larvae.
Gosner, K. L. 1960. A simplified table for staging anuran
embryos and larvae with notes on identification. Herpeto-
Gray, H. M., M. Ouellet, D. M. Green, and A. S. Rand. 2002.
Traumatic injuries in two Neotropical frogs, Dendrobates
auratus and Physalaemus pustulosus. Journal of Herpetology
Guderyahn, L. 2006. Nationwide assessment of morphological
abnormalities observed in amphibians collected from United
States National Wildlife Refuges. CBFO-C0601. U.S. Fish
and Wildlife Service, Annapolis, Maryland, USA.
Halekoh, U., S. Højsgaard, and J. Yan. 2001. The R Package
geepack for generalized estimating equations. Journal of
Statistical Software 15:1–11.
Horton, N. J., and S. R. Lipsitz. 1999. Review of software to fit
generalized estimating equation regression models. American
Johnson, P. T. J., K. B. Lunde, R. W. Haight, J. Bowerman,
and A. R. Blaustein. 2001a. Ribeiroia ondatrae (Trematoda:
Digenea) infection induces severe limb malformations in
western toads (Bufo boreas). Canadian Journal of Zoology
Johnson, P. T. J., K. B. Lunde, E. G. Ritchie, and A. E. Launer.
1999. The effect of trematode infection on amphibian limb
development and survivorship. Science 284:802–804.
Johnson, P. T. J., K. B. Lunde, E. G. Ritchie, J. K. Reaser, and
A. E. Launer. 2001b. Morphological abnormality patterns in a
California amphibian community. Herpetologica 57:336–352.
Johnson, P. T. J., K. B. Lunde, E. M. Thurman, E. G. Ritchie,
S. N. Wray, D. R. Sutherland, J. M. Kapfer, T. J. Frest, J.
Bowerman, and A. R. Blaustein. 2002. Parasite (Ribeiroia
ondatrae) infection linked to amphibian malformations in the
western United States. Ecological Monographs 72:151–168.
Johnson, P. T. J., E. R. Preu, D. R. Sutherland, J. Romansic, B.
Han, and A. R. Blaustein. 2006. Adding infection to injury:
synergistic effects of predation and parasitism on salamander
limb malformations. Ecology 87:2227–2235.
Johnson, P. T. J., M. K. Reeves, S. K. Krest, and A. E.
Pinkney. In press. A decade of deformities: advances in our
understanding of amphibian malformations and their impli-
cations. In D. W. Sparling, G. Linder, C. A. Bishop, and
S. K. Krest, editors. Ecotoxicology of amphibians and
reptiles. Second edition. Society of Environmental Toxicol-
ogy and Chemistry, Pensacola, Florida, USA.
Kaliszewicz, A. 2003. Sublethal predation on Stylaria lacustris:
a study of regenerative capabilities. Hydrobiologia 501:83–
JAY BOWERMAN ET AL.250 Ecology, Vol. 91, No. 1
Kiesecker, J. M. 2002. Synergism between trematode infection Download full-text
and pesticide exposure: A link to amphibian limb deformities
in nature? Proceedings of the National Academy of Sciences
Kurabuchi, S., and S. Inoue. 1982. Limb regenerative capacity
of four species of Japanese frogs of the families Hylidae and
Ranidae. Journal of Morphology 173:129–135.
Lannoo, M. J. 2008. Malformed frogs: the collapse of aquatic
ecosystems. University of California Press, Berkeley, Cal-
Lannoo, M. J., D. R. Sutherland, P. Jones, D. Rosenberry,
R. W. Klaver, D. M. Hoppe, P. T. J. Johnson, K. B. Lunde,
C. Facemire, and J. M. Kapfer. 2003. Multiple causes for the
malformed frog phenomenon. Page 1443 in G. Linder, E.
Little, S. Krest, and D. Sparling, editors. Multiple stressor
effects in relation to declining amphibian populations.
ASTM International, West Conshoshocken, Pennsylvania,
Lawler, S. P., D. Dritz, T. Strange, and M. Holyoak. 1999.
Effects of introduced mosquitofish and bullfrogs on the
threatened California red-legged frog. Conservation Biology
Liang, K. Y., and S. Zeger. 1986. Longitudinal data analysis
using generalized linear models. Biometrika 73:13–22.
Meteyer, C. U., I. K. Loeffler, J. F. Fallon, K. A. Converse,
D. E. Green, J. C. Helgen, S. Kersten, R. Levey, L. Eaton-
Poole, and J. G. Burkart. 2000. Hind limb malformations in
free-living northern leopard frogs (Rana pipiens) from Maine,
Minnesota, and Vermont suggest multiple etiologies. Tera-
Meyer, J. J., and J. E. Byers. 2005. As good as dead? Sublethal
predation facilitates lethal predation on an intertidal clam.
Ecology Letters 8:160–166.
Mouritsen, K. N., and R. Poulin. 2003. The risk of being at the
top: foot-cropping in the New Zealand cockle Austrovenus
stutchburyi. Journal of the Marine Biological Association of
the United Kingdom 83:497–498.
Ouellet, M. 2000. Amphibian deformities: current state of
knowledge. Pages 617–661 in D. W. Sparling, G. Linder, and
C. A. Bishop, editors. Ecotoxicology of amphibians and
reptiles. Society of Environmental Toxicology and Chemis-
try, Pensacola, Florida, USA.
Pape-Lindstrom, P. A., R. J. Feller, S. E. Stancyk, and S. A.
Woodin. 1997. Sublethal predation: field measurements of
arm tissue loss from the ophiuroid Microphiopholis gracillima
and immunochemical identification of its predators in North
Inlet, South Carolina, USA. Marine Ecology Progress Series
Reeves, M. K., C. L. Dolph, H. Zimmer, R. S. Tjeerdema, and
K. A. Trust. 2008. Road proximity increases risk of skeletal
abnormalities in wood frogs from National Wildlife Refuges
in Alaska. Environmental Health Perspectives 116:1009–
Schoener, T. W. 1979. Inferring the properties of predation and
other injury-producing agents from injury frequency. Ecol-
Sessions, S. K., R. A. Franssen, and V. L. Horner. 1999.
Morphological clues from multilegged frogs: Are retinoids to
blame? Science 284:800–802.
Sessions, S. K., and S. B. Ruth. 1990. Explanation for naturally
occurring supernumerary limbs in amphibians. Journal of
Experimental Zoology 254:38–47.
Skelly, D. K., S. R. Bolden, L. K. Freidenburg, N. A.
Freidenfelds, and R. Levey. 2007. Ribeiroia infection is not
responsible for Vermont amphibian deformities. EcoHealth
Smith, D. C. 1983. Factors controlling tadpole populations of
the chorus frog (Pseudacris triseriata) on Isle Royale,
Michigan. Ecology 64:501–510.
Souder, W. 2000. A plague of frogs. Hyperion, New York, New
Stopper, G. F., L. Hecker, R. A. Franssen, and S. K. Sessions.
2002. How trematodes cause limb deformities in amphibians.
Journal of Experimental Zoology 294:252–263.
Stuart, S. N., J. S. Chanson, N. A. Cox, B. E. Young, A. S. L.
Rodrigue, D. L. Fischman, and R. W. Waller. 2004. Status
and trends of amphibian declines and extinctions worldwide.
Van Buskirk, J., and S. A. McCollum. 2000. Influence of tail
shape on tadpole swimming performance. Journal of
Experimental Biology 203:2149–2158.
Van Gelder, J. J., and H. Strijbosch. 1995. Adult common toads
(Bufo bufo) with mutilated legs. Alytes 13:105–108.
Veith, M., and B. Viertel. 1993. Vera ¨ nderungen an den
extremita ¨ ten von Larven und Jungtieren der erdkro ¨ te (Bufo
bufo): Analyse mo ¨ glicher Ursachen. Salamandra 29:184–199.
Viertel, B., and M. Veith. 1992. Predation by leeches and
regeneration, a factor in larval development of Bufo bufo (L.).
Pages 479–484 in Z. Korsos and I. Kiss, editors. Proceedings
of the Sixth Ordinary General Meeting of the Societas
Europaea Herpetologica, Budapest, 1991. Hungarian Natu-
ral History Museum, Budapest, Hungary.
Wilbur, H. M., and J. E. Fauth. 1990. Experimental aquatic
food webs: interactions between two predators and two prey.
American Naturalist 135:176–204.
Wildy, E. L., D. P. Chivers, J. M. Kiesecker, and A. R.
Blaustein. 2000. The effects of food level and conspecific
density on biting and cannibalism in larval long-toed
salamanders, Ambystoma macrodactylum. Oecologia 128:
Wisniewiski, W. L. 1958. Characterization of the parasitofauna
of a eutrophic lake. Acta arasitologica Polonica 6:1–64.
January 2010251PREDATION AND ABNORMAL AMPHIBIANS