Content uploaded by Heather Gray
Author content
All content in this area was uploaded by Heather Gray on Apr 29, 2014
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Phyllomedusa
- 8(1), July 2009
41
Kin discrimination in cannibalistic tadpoles of the
Green Poison Frog, Dendrobates auratus (Anura,
Dendrobatidae)
Heather M. Gray1, Kyle Summers2,3 and Roberto Ibáñez D.3
1Redpath Museum, McGill University, 859 Sherbrooke St. W., Montreal, Quebec, Canada H3A 2K6. E-mail:
grayhm@hotmail.com.
2Department of Biology, East Carolina University, Greenville, NC 27834, USA. E-mail: summersk@mail.ecu.edu.
3Smithsonian Tropical Research Institute, Attn: Roberto Ibáñez D., Apartado 2072, Balboa, Ancón, Republica de Panamá.
E-mail: ibanezr@tivoli.si.edu.
Received 5 December 2008.
Accepted 23 April 2009.
Distributed July 2009.
Phyllomedusa 8(1):41-50, 2009
© 2009 Departamento de Ciências Biológicas - ESALQ - USP
ISSN 1519-1397
Abstract
Kin discrimination in cannibalistic tadpoles of the Green Poison Frog,
Dendrobates auratus (Anura, Dendrobatidae). Cannibalizing a related individual
can reduce the inclusive fitness of the cannibal. Hence, mechanisms that allow a
tadpole to recognize and modify its behavior toward kin may reduce the inclusive
fitness costs of cannibalism. Alternatively, ecological factors may cause preferential
treatment of kin to be too costly to be favored by selection. We tested these two
predictions in the Green Poison Frog, Dendrobates auratus. The effect of kinship
on larval cannibalism was examined through a series of kin-discrimination trials.
The behavior of large tadpoles was observed when presented with two small,
tethered tadpoles, one a clutchmate and one an unrelated tadpole. In these
simultaneous presentation tests, tadpoles displayed a significant preference for
attacking kin. In a series of timed trials, pairs of unequally sized tadpoles were
placed together in containers. The majority (70%) of large tadpoles took less than
24 hr to consume the small tadpole. Kinship did not affect the survival time of the
small tadpole. Our results are consistent with observations that D. auratus is an
indiscriminate predator. As conspecifics may be serious competitors, their swift
elimination would be an advantage, particularly in the small, nutrient-poor pools
used by this species.
Keywords: Anura, Dendrobatidae, cannibalism, kinship, larvae.
Phyllomedusa
- 8(1), July 2009
42
Introduction
Hamilton’s inclusive fitness theory provides
a framework to understand the evolution of
altruistic and aggressive behavior (Hamilton
1964). Hamilton’s Rule predicts that altruism
(or reduced aggression) will be favored when
rb – c > 0, where c is the fitness cost to the
altruist, b is the fitness benefit to the beneficiary
and r is their genetic relatedness (Hamilton
1964). The rationale behind preferential
treatment of related individuals is that it will aid
the propagation of genes shared with kin.
To behave differently toward kin, related
individuals must be recognized as such
(Hamilton 1964). Kin recognition has been
identified in a wide variety of organisms,
including many amphibians (reviewed in
Fletcher and Michener 1987, Sherman et al.
1997). Studies of kin recognition in amphibians
have often examined it in the context of
cannibalism (Walls and Roudebush 1991,
Pfennig and Collins 1993, Pfennig et al. 1993,
Pfennig 1997). Cannibalism, or the killing and
ingestion of conspecifics, is widespread in
nature (Polis 1981, Elgar and Crespi 1992) and
is found in many groups of amphibians, with
the most frequent type being that of larvae
Resumo
Discriminação de relativos em girinos canibais de Dendrobates auratus (Anura, Dendrobatidae).
O consumo de indivíduos aparentados pode reduzir a aptidão inclusiva do canibal. Assim, mecanismos
que permitam que um girino reconheça seus relativos e modifique seu comportamento pode reduzir
os custos do canibalismo. Alternativamente, fatores ecológicos podem tornar o tratamento preferencial
dos relativos custoso demais para que seja favorecido pela seleção natural. Testamos essas duas
previsões no dendrobatídeo Dendrobates auratus. O efeito do parentesco sobre o canibalismo larval
foi examinado por meio de uma série de tentativas de discriminação de parentes. Observamos o
comportamento de girinos de grande porte diante de dois girinos menores imobilizados, um irmão e
outro não-aparentado. Nesses testes de apresentação simultânea, os girinos mostraram uma preferência
significativa por atacar irmãos. Em outra série de tentativas, pares de girinos de tamanhos diferentes
foram colocados juntos em aquários. A maioria dos girinos de grande porte (70%) consumiu o girino
menor em menos de 24 horas. O parentesco não afetou o tempo de sobrevivência do girino pequeno.
Nossos resultados são consistentes com as observações de que D. auratus é um predador
indiscriminado. Como os co-específicos podem ser fortes competidores, sua eliminação rápida poderia
ser vantajosa, particularmente nas pequenas poças pobres em nutrientes utilizadas por essa espécie.
Palavras-chave: Anura, Dendrobatidae, canibalismo, parentesco, larvas.
consuming larvae (Crump 1992). The benefits
derived from consuming conspecifics are
primarily nutritional, allowing the cannibal
enhanced growth or survival success (Crump
1992). There may be additional benefits of
consuming conspecifics as they represent the
proper proportion of nutrients required for the
cannibal’s growth (Nagai et al. 1971, Meffe and
Crump 1987, Crump 1990, Wildy et al. 1998).
Another benefit of cannibalism may be that it
eliminates sources of competition (Polis 1981).
Despite the nutritional benefits of
cannibalism, there are costs. These include the
risk of injury to the attacking animal, greater
exposure to species-specific disease and the risk
of eating a relative (Polis 1981, Crump 1992).
Cannibals tend to attack animals smaller than
themselves or animals that are unable to defend
themselves (Elgar and Crespi 1992), thereby
reducing the risk of injury. More efficient
spread of pathogens to conspecifics has been
demonstrated in a variety of organisms (Polis
1981, Pfennig et al. 1991) and has been
suggested as the main reason certain animals
avoid cannibalism (Pfennig et al. 1998). The
loss of inclusive fitness by eating a relative can
have a direct selective impact (Hamilton 1964).
A number of amphibian species reduce this
Gray et al.
Phyllomedusa
- 8(1), July 2009
43
impact by avoiding cannibalizing kin (examples
of anurans: Bufo americanus [Waldman 1985],
Spea bombifrons and S. multiplicata [Pfennig
and Frankino 1997], and salamanders: Hynobius
retardatus [Wakahara 1997], Ambystoma
tigrinium nebulosum [Pfennig and Collins 1993,
Pfennig et al. 1999]). In contrast, some species
do not seem to avoid cannibalizing kin (e.g.,
Walls and Blaustein 1995). The evolution of
mechanisms of kin discrimination should
depend on how larval ecology affects the costs
and benefits of cannibalism.
The Neotropical anuran genus Dendrobates
comprises many species with larval ecologies
that make cannibalism advantageous. Most
Dendrobates tadpoles have massive beaks with
serrated teeth that would allow carnivory
(Silverstone 1975, Caldwell and Myers 1990).
Combined with this morphogical modification,
the aggressive behavior of Dendrobates
tadpoles renders them successful predators
(Summers 1990, Caldwell and Araújo 1998,
Summers 1999). The exceptions to this are the
obligate egg-eating tadpoles of the Oophaga
histrionicus Group that survive on unfertilized
trophic eggs provided by the female parent
(Weygoldt 1987), and therefore these tadpoles
are excluded from further discussion.
Dendrobates deposit terrestrial eggs that
develop into tadpoles, and then are transported
by a parent to a pool of water to complete
development. Females produce many small
clutches of eggs year round so there is a steady
supply of tadpoles to be deposited (Weygoldt
1987). All Dendrobates tadpoles are deposited
in small pools of water (i.e., phytotelmata),
which include water that collects in fruit husks,
such as capsules of Brazil nuts (Caldwell 1993),
calabash (Summers 1990) and cacao pods
(H. M. Gray, pers. observ.), tree holes (Dunn
1941, Eaton 1941) and leaf axils of plants
(Summers 1992, Brust 1993). Not only are
these pools small and at risk of drying out, but
often they are freshly formed the night before
tadpole deposition, and thus are low in accumu-
lated nutrients (H. M. Gray, pers. observ.).
For Dendrobates auratus, the benefits
derived from consuming conspecifics are
primarily nutritional, allowing the cannibal
enhanced growth and survival (Crump 1992).
D. auratus tadpoles are generally voracious
predators and will consume any insect larvae,
and heterospecific or conspecific tadpoles
deposited in the same pool (Summers 1990,
Caldwell and Araújo 1998). Conspecific
tadpoles provide a relatively large prey item in
a usually nutrient-poor environment. Increasing
the food available to D. auratus tadpoles
increases larval growth rate (Summers 1990).
Although the consumption of conspecifics has
not been shown to confer any additional
advantage relative to other sources of calories
and nutrients, cannibalism does eliminate the
tadpoles’ closest competitors, as well as their
potential predators (i.e., other cannibals).
The costs of cannibalism in Dendrobates
auratus include the risk of injury to the
attacking animal, pathogen transmission, and
the loss of inclusive fitness (Crump 1992, Polis
1981). The risk of injury to tadpoles attempting
to cannibalize conspecifics is apparently high;
in pools in which there are more than two D.
auratus, individuals have damaged tails
(Summers 1990). Tadpoles usually are able to
complete metamorphosis despite tail damage
(H. M. Gray, unpubl. data), although these
injuries may impose a cost in terms of reduced
growth and developmental rates as was found
in Rana catesbeiana (Wilbur and Semlitsch
1990). Pathogen transmission via cannibalism
in this species has not been explored. Also, it
is unknown whether any mechanism exists to
recognize kin and avoid a loss of inclusive
fitness in D. auratus.
Field observations indicate that individual
male Dendrobates auratus return to pools and
deposit tadpoles from different clutches into the
same pool, although they typically deposit
tadpoles from a single clutch into different
pools (Summers 1990). In most pools where
tadpoles were deposited, more than one male
was observed bringing tadpoles to the pool
Kin discrimination in cannibalistic tadpoles of the Green Poison Frog
Phyllomedusa
- 8(1), July 2009
44
(Summers 1990). Hence, pools typically contain
a mixture of related and unrelated tadpoles.
Because tadpoles are rarely deposited
simultaneously, they typically vary in size.
Even under experimental conditions when
tadpoles are deposited at similar sizes, they
rapidly diverge in size, owing to intraspecific
competition (Summers 1990). Typically, the
total number of tadpoles deposited is small;
thus, cannibals do not have the option of
choosing among a large number of relatives and
non-relatives.
We examined tadpole choice by simulta-
neously presenting a large tadpole with a
small related and a small unrelated
conspecific. We also examined, in a series of
paired timed trials, the time for a large
tadpole to cannibalize a small tadpole, and
whether kinship influences the outcome. The
simplest prediction with regard to kin
discrimination in Dendrobates auratus is that
tadpoles will avoid eating relatives, thereby
increasing their inclusive fitness. However,
owing to the larval ecology of D. auratus, this
prediction may be too simplistic. Selection
may not favor kin discrimination if the small
size and nutrient-poor status of the pools, and
the associated high levels of competition and
cannibalism, make kin discrimination
prohibitively costly.
Materials and Methods
Tadpole collection and maintenance
Experiments were conducted during May–
July 1998 and May–August 1999 in the
Republic of Panama. Sibling Dendrobates
auratus tadpoles were obtained from two
sources. Tadpoles were collected from the
backs of males transporting more than one
tadpole. Males usually transport a single
tadpole, but occasionally transport more.
During a 2-mo period, the number of tadpoles
being transported by males was counted at the
two field sites (Cerro Ancon, 8°56’N,
79°34’W and Isla Taboga, 8°47’N, 79°34’W).
Of 46 males observed transporting tadpoles,
42 (91.3%) were transporting a single
tadpole, 3 (6.52%) were carrying two, and 1
individual (2.17%) was carrying three
tadpoles. Males were assumed to be the father
of the tadpoles that they carried. Cuckoldry
of one male by another has not been observed
in field studies of the mating system of D.
auratus, and the aggressive territoriality of
males in this species makes cuckoldry
unlikely (Summers 1989). Because multiple
tadpoles on a given male’s back were always
of the same size, they were assumed to be of
the same clutch.
The second source of tadpoles was a group
of captive Dendrobates auratus. Breeding pairs
of adults were captured in the field and were
maintained in 10-gal (37.85-L) terraria at the
Tupper Center of the Smithsonian Tropical
Institute in Panama City. Terraria were kept at
ambient temperature and were furnished with
leaf litter, a hollowed branch, a bowl of
dechlorinated water, and a dish containing
pieces of ripe fruit. Animals were fed lab-
reared, wingless Drosophila dusted with
vitamin powder. They were also able to eat the
insects that were attracted to the rotting fruit in
the terraria. Tadpoles were collected from the
water bowls placed in the terraria with a single
pair of frogs. The date tadpoles were collected
was noted in order to keep track of which
tadpoles were from the same clutch. All related
tadpoles used were full siblings. In total, 29
sibships were used in the experiments. Nineteen
different sibships were examined for the kin
discrimination trials, and 10 different sibships
were used in the timed trials. Experiments were
conducted with tadpoles that were 7–14 days
old.
Tadpoles to be used as the unrelated
individuals were collected from the field in
areas known to have large breeding populations
of Dendrobates auratus. Small plastic cups,
filled with dechlorinated water, were placed in
areas not overlapping with the source areas for
Gray et al.
Phyllomedusa
- 8(1), July 2009
45
the related tadpoles. The next day, all tadpoles
deposited in these cups were collected.
Tadpoles from all sources were maintained
individually in plastic cups containing 75 ml of
dechlorinated water and were fed commercial
goldfish food every few days.
Experiment 1: Kin discrimination trials
Each kin discrimination trial involved three
tadpoles—i.e., a pair of siblings and one
unrelated individual. One of the siblings was
designated the test tadpole. This animal was fed
fish food and mosquito larvae ad libitum for the
week prior to trials. The other tadpoles were fed
limited fish food only. The trials were
conducted with unequal-sized individuals
because this allowed the experiment to progress
more rapidly. In the field, when individuals of
the same size share a small pool, one will
outcompete the others and will increase in size
much more rapidly than the others (Summers
1990).
The day before the trial, tadpoles were
weighed (Salter AND electronic balance FX-
200) and their body length from the snout to the
junction of the posterior body wall with the tail,
or snout-vent length (SVL), measured with dial
calipers. In 1998, only the small tadpoles were
measured, whereas in 1999, the test tadpoles
also were measured. Each small kin tadpole was
carefully matched for size with an unrelated
tadpole. All three tadpoles were placed into
separate containers with clean water and left
without food for 24 hr to ensure that the animals
were all at the same hunger level during trials.
The experimental arena (Figure 1) was a
clear plastic rectangular tank (18.5 × 13.25 cm).
A black line was drawn on the bottom of the
tank dividing it into halves. The tank was filled
with water to a depth of 2.5 cm. We placed an
open-ended net cylinder in the middle of the
tank, straddling the centerline. The cylinder was
a tube of fine open-meshed material (tulle) with
wire rings (5-cm diameter) at either end to give
the tube rigidity. The test tadpole was placed in
this cylinder during acclimation to the arena. A
short piece of black thread was tied around the
base of the tail of each of the small tadpoles.
The end of the thread of each tethered tadpole
was then taped to opposite sides of the test
arena. In this way, the small tadpoles were able
to swim freely around in their half of the test
arena but were unable to cross the center line.
The side of the tank occupied by the related or
unrelated tadpole was switched for each trial.
After a 20-min acclimation period, the cylinder
was removed and the test tadpole was able to
swim freely about the entire arena. Interactions
between the test tadpole and its smaller sibling
and the smaller unrelated tadpole were filmed
for 1 hr with a video camera (Sony Hi-8 video
CCD-TR100 NTSC). At the end of the hour, the
small tadpoles were untied and their condition
noted. The videotapes were then watched by an
observer who noted the number of attacks and
bites the test tadpole inflicted on each of the
smaller animals. To control for any a priori
assumptions about outcome, the observer did
not know the relatedness of the tadpoles when
scoring the videotapes.
Figure 1 - Experimental arena during acclimation period.
After 20-min of acclimation, the net cylinder
was removed and the interactions between the
test tadpole and its clutch mate and an
unrelated tadpole (shaded grey on the right
side of the tank) were filmed. Tethering
ensured that the small tadpoles remained on
opposite sides of the tank.
Kin discrimination in cannibalistic tadpoles of the Green Poison Frog
Phyllomedusa
- 8(1), July 2009
46
Experiment 2: Timed cannibalism trials
To determine if relatedness of tadpoles
affected the amount of time for a large tadpole
to consume a small tadpole, we conducted
paired trials with animals in small containers.
These experiments also controlled for the effect
of tethering in the first experiment, ensuring
that the attacks on the small tadpoles in the first
experiment did not occur simply because the
tadpoles were tethered. Two tadpoles, again one
larger than the other, were placed together in
small plastic containers containing 75 mL of
dechlorinated water. Each paired trial consisted
of a container holding a large and small, related
tadpole and a container holding a large and
small, unrelated tadpole. In each set of trials,
large and small animals were carefully matched
for size, both in wet weight and snout-vent
length. After measuring, all tadpoles were
placed individually into clean containers with
water and left for the 24 hr prior to the
experiment. After the 24 hr, the large and
small, related animals were put together and
the large and small, unrelated animals were
put together in a separate container. A pellet
of goldfish food was added to each container
with the pair of tadpoles. This was done to
give the big tadpole a choice of foods and
eliminate the possibility that a starving
tadpole would indiscriminately eat any
tadpole in the same pool. Containers were
checked every 12 hr and the time taken for
the large tadpole to kill the small tadpole was
noted.
Statistical analysis
Statistical analyses were carried out using
Systat Version 8.0 (SPSS Inc., 1998). Data were
tested for normality of distribution
(Kolmogorov-Smirnov normality test) and for
homogeneity of variances (Bartlett’s test).
Proportional data were arcsine–square root
transformed to meet parametric assumptions of
normality (Zar 1999).
In experiment 1, body length and mass
differences were examined with two-sample t-
tests. Kruskal-Wallis one-way analysis of
variance was used to determine if the test
tadpoles were significantly larger than the
tethered tadpoles. To determine if the number
of attacks and bites were based on kinship, a
Mann-Whitney U-test was used. To take into
consideration the range in attack behavior of the
different test tadpoles, we analyzed, with t-tests,
the proportion of attacks and bites inflicted on
related and unrelated small tadpoles.
In experiment 2, body length and mass
differences were examined with two-sample t-
tests. To determine if kinship influenced the
time to cannibalism, a Mann-Whitney U-test
was used (Zar 1999).
Results
Experiment 1: Kin discrimination trials
There were no significant differences in the
sizes of the small tadpoles used in each kin
discrimination trial (mean ± standard deviation;
small tadpoles related to the test tadpole
0.065 ± 0.0184 g wet weight, 6.5 ± 0.704 mm
body length, N = 19; small tadpoles unrelated
to the test tadpole 0.064 ± 0.0165 g,
6.49 ± 0.603 mm, N = 19; two-sample t-test for
mass: t36 = 0.315, P = 0.754; two-sample t-test
for body length: t36 = 0.025, P = 0.98). The test
tadpoles were significantly larger than the small
tadpoles (0.12 ± 0.0435 g wet weight,
7.97 ± 0.986 mm body length, N = 13; Kruskal-
Wallis one-way analysis of variance for mass:
H = 16.11, k = 3, P << 0.05; analysis of
variance for SVL: F = 15.94, df(factor) = 2,
df(residual) = 36, P << 0.05). The test tadpoles
were almost twice the weight and on average
1.5 mm longer than the small tadpoles.
There was no significant difference in the
number of attacks on the small tadpoles based
on kinship (mean number of attacks on kin
8.89 ± 9.73 and on nonkin 4.63 ± 6.73, Mann-
Whitney U-test: U = 235, N1 = N2 = 19, P
Gray et al.
Phyllomedusa
- 8(1), July 2009
47
= 0.104). Similarly, there was no significant
difference in the number of bites inflicted by the
test tadpole on the small related or unrelated
tadpoles (mean number of bites to kin
49.63 ± 159.831 and to nonkin 11 ± 16.9345;
Mann-Whitney U-test: U = 224, N1 = N2 = 19,
P = 0.195). The large variance in numbers of
bites was attributable, in particular, to one test
tadpole that attacked the related small tadpole
23 times, biting the small tadpole 707 times.
This value is an order of magnitude greater than
the next highest number of bites (65). The test
tadpoles displayed a significant preference for
attacking kin (proportion of attacks on kin,
0.74 ± 0.302; one sample t-test: t15 = 2.821,
P = 0.013). There was no significant preference,
relative to a baseline of 50%, found when the
proportion of bites inflicted was examined
(proportion of bites to kin 0.676 ± 0.358; one
sample t-test: t15 = 1.670, P = 0.116). There
were however significantly greater proportions
of both attacks and bites made to kin tadpoles
when compared to the proportion of attacks and
bites to nonkin (two-sample t-test for attacks:
t30 = 3.99, P = 0.000; two-sample t-test for
bites: t30 = 2.361, P = 0.025).
The use of the observed condition of the
tadpoles as they were removed from the setup
proved problematic as the test tadpole often
continued to attack tadpoles during the
dismantling of the arena. The end result was
that during the 19 trials, three kin and three
unrelated small tadpoles were fatally wounded
by the time they were removed from the test
arena. From the videotapes, only the three
related and one of the unrelated small tadpoles
were attacked during filming meaning that
enough injuries were sustained during the
dismantling period to kill an additional two
unrelated tadpoles.
There was no significant tendency for the
test tadpole to preferentially attack or bite the
tadpole on a particular side of the test arena
(one-sample t-test for attacks: t15 = –1.833,
P = 0.087; one-sample t-test for bites: t15= –
1.796, P = 0.093).
Experiment 2: Timed cannibalism trials
The large tadpoles in the timed kin trials
(0.14 ± 0.038 g; 8.29 ± 0.887 mm, N = 10) and
the large tadpoles in the timed unrelated trials
(0.14 ± 0.0370 g; 8.49 ± 1.01 mm, N = 10) were
not significantly different in size (two-sample
t-test for mass: t18 = 0.254, P = 0.802; two-
sample t-test for body length: t18 = –0.470,
P = 0.644). There was also no significant
difference in size of the small tadpoles (small
tadpoles in kin trials 0.061 ± 0.0147 g,
6.41 ± 0.499 mm body length, N = 10; small
tadpoles in unrelated trials 0.061 ± 0.0141 g,
6.22 ± 0.657 mm, N = 10; two-sample t-test for
mass: t18 = –0.015, P = 0.988; two-sample t-test
for body length: t18 = 0.727, P = 0.476). The
large tadpoles were significantly larger than the
small tadpoles (two-sample t-test for mass:
t38 = 8.71, P = <<0.05; two-sample t-test for
body length: t38 = 8.47, P = <<0.05).
It took large tadpoles less than 12 hours to
up to two weeks to fully consume the smaller
tadpole in the container. The majority of large
tadpoles took less than 24 hours to consume the
smaller tadpole (70%, N = 20). In four of 10
paired trials, the time that it took for the large
tadpole to kill the small tadpole was the same.
In three trials, the small tadpole in the related
container died first and in three trials the
unrelated small animal died first. There was no
significant difference between the mean time
until cannibalism between trials with kin and
trials with non-kin (Mann-Whitney U test,
N = 10, Xkin = 42.8 hr, SE = 14.6 hr, Xnonkin
= 22.8 hr, SE = 3.3 hr, U = 41, P = 0.496).
Discussion
In some species, tadpoles recognize their kin
and modify their cannibalistic behavior in order
to avoid a reduction of inclusive fitness (Crump
1990). We found no evidence that Dendrobates
auratus tadpoles avoid cannibalizing their kin.
Our results demonstrated that the test tadpoles
directed a higher proportion of both attacks and
Kin discrimination in cannibalistic tadpoles of the Green Poison Frog
Phyllomedusa
- 8(1), July 2009
48
bites toward related, rather than unrelated
tadpoles, although tadpoles in both groups
ultimately suffered the same fate. The numbers
of attacks and bites directed at unrelated
tadpoles were not significantly different than
those directed at related tadpoles, and an equal
number of unrelated and related small tadpoles
ultimately died in the preference trials. In
experiment 2, the similar speed with which
large tadpoles killed and consumed unrelated or
related smaller tadpoles confirms that even if
tadpoles are able to discriminate kin from
unrelated tadpoles this does not modify their
cannibalistic behavior.
Our results raise several questions. First,
why did large tadpoles not discriminate in favor
of their kin, as has been observed in other
species? Most anuran species that possess kin
discrimination live in temporary or permanent
ponds (e.g., Spea bombifrons, Pfennig et al.
1993). Because these ponds contain large
clutches of related and unrelated larvae, there
are typically many individuals from which to
choose and a cannibal can be selective and
avoid eating a relative. This is not the case for
Dendrobates auratus tadpoles which are
deposited into small, nutrient-poor pools.
Indiscriminate killing of related and unrelated
conspecifics occurs in other species that utilize
small pools (e.g., Adelphobates castaneoticus
[Caldwell and Araújo 1998] and mosquito
larvae [Sheratt et al. 1999]). In such pools,
competition for severely limited resources may
make the costs of altruism toward kin (e.g.,
refraining from cannibalism) too high (Pfennig
1997). Given the small size of the pools utilized
by D. auratus, there may simply not be enough
food available for a tadpole to allow even a
small full sibling to survive and compete for
resources. Hence, the best strategy may be to
eat all small tadpoles introduced into the pools.
It is possible that cannibalism in
Dendrobates auratus occurs as a result of
completely indiscriminate predation, in which
all competitors and potential predators are
attacked, and no distinction is made between
different species (Crump 1992, Caldwell and
Araújo 1998). D. auratus tadpoles are efficient
predators of the tadpoles of sympatric species
of frogs (Summers 1990, Caldwell and Araújo
1998). D. auratus tadpoles also attack
invertebrate larvae (Finke 1994). Other than
some of these larvae which become predators
when large, the greatest threat to D. auratus
tadpoles are conspecifics, which may explain
why the large tadpoles in experiment 2 chose
to kill and consume the smaller tadpole in
preference to eating the fish food. The drive to
attack was not simply a consequence of hunger,
implying that the calories and nutrients gained
from cannibalism are not the immediate
motivator of the behavior. Instead, the
elimination of a potentially long-term
competitor (or predator) may be the most
important benefit.
A second reason that large tadpoles did not
discriminate in favor of kin may be that the risk
of attacking close relatives is comparatively
low. The tadpoles of Dendrobates auratus are
transported individually by their male parent to
a pool. The male chooses where to deposit each
of his offspring. Male D. ventrimaculatus have
been shown to avoid placing young into pools
which already contain a large tadpole (Summers
1999), and the same may be true for D. auratus.
Male D. auratus spend substantial amounts of
time evaluating and searching for suitable pools
to deposit tadpoles (Summers 1990). Part of this
evaluation may involve the number and the size
of other tadpoles, as well as the presence of
previous offspring in the pool. Males transport
tadpoles of the same clutch to different pools
(Summers 1990). These deposition strategies of
male D. auratus may reduce the probability that
tadpoles will encounter and consume close
relatives, and hence decrease the indirect fitness
costs of indiscriminate predation of
conspecifics.
Another question is why did kin receive a
greater proportion of attacks? One possibility is
that tadpoles were able to recognize kin as
potential competitors more quickly and
Gray et al.
Phyllomedusa
- 8(1), July 2009
49
efficiently than they were able to recognize
non-kin. It is not known what cue tadpoles use
to determine either relatedness or conspecific
status. Other species of amphibians have been
shown to use chemical cues (olfactory) to
recognize kin (Pfennig and Collins 1993). Adult
Dendrobates pumilio have been shown to use
olfactory cues in territory identification
(Forester and Wisnieski 1991), and tadpoles of
Dendrobates species may also have well-
developed chemosensory reception. The D.
auratus tadpoles may be using olfactory cues to
identify another tadpole in the pool. As the
experimental tadpoles were raised in isolation,
the test tadpole may recognize a tadpole that
smells like itself as a conspecific competitor
more quickly and this may explain the initial
preference for kin. This use of self to develop
a standard for comparison is described by
Grafen (1990), who posited that kin recognition
(and discrimination) might be an artifact of
species recognition. Specifically, he proposed
that, if the cues used for species recognition are
perceived by matching those cues to a template
derived from the animals’ own phenotype, then
individuals may tend to recognize closely
related individuals more effectively, simply as
a byproduct of the recognition system. This
could produce non-adaptive biases in
recognition and discrimination that are
associated with kinship.
Acknowledgements
We thank the scientific and administrative
staff of the Smithsonian Tropical Research
Institute for logistic support and advice, and the
Autoridad Nacional Del Ambiente (ANAM) of
the Republic of Panama for collection and
research permits. Permission to carry out the
research was obtained from ANAM and from
the animal care and use committee of East
Carolina University. We thank A.S. Rand, S.
MacKenzie, D. Pfennig, and J. Stamps for
comments and discussion. The line drawing was
kindly done by A.V. Savage.
References
Brust, D. G. 1993. Maternal brood care by Dendrobates
pumilio: a frog that feeds its young. Journal of
Herpetology 27: 96–98.
Caldwell, J. P. 1993. Brazil nut fruit capsules as
phytotelmata: interactions among anuran and insect
larvae. Canadian Journal of Zoology 71: 1193–1201.
Caldwell, J. P. and M. C. Araújo. 1998. Cannibalistic
interactions resulting from indiscriminate predatory
behavior in tadpoles of the poison frogs (Anura:
Dendrobatidae). Biotropica 30: 92–103.
Caldwell, J. P. and C. W. Myers. 1990. A new poison frog
from Amazonian Brazil, with further revision of the
quinquevittatus group of Dendrobates. American
Museum Novitates 2988: 1–21.
Crump, M. L. 1990. Possible enhancement of growth in
tadpoles through cannibalism. Copeia 1990: 560–564.
Crump, M. L. 1992. Cannibalism in amphibians. Pp. 256–
276 in M. A. Elgar and B. J. Crespi (eds.),
Cannibalism: Ecology and Evolution Among Diverse
Taxa. Oxford. Oxford University Press.
Dunn, E. R. 1941. Notes on Dendrobates auratus. Copeia
1941: 88–93.
Eaton, T. H. Jr. 1941. Notes on the life history of
Dendrobates auratus. Copeia 1941: 93–95.
Elgar, M. A. and B. J. Crespi (eds.). 1992. Cannibalism:
Ecology and Evolution Among Diverse Taxa. Oxford.
Oxford University Press. 376 pp.
Finke, O. M. 1994. Population regulation of a tropical
damselfly in the larval stage by food limitation,
cannibalism, intraguild predation and habitat drying.
Oecologia 100: 118–127.
Fletcher, D. J. C and C. D. Michener. 1987. Kin
Recognition in Animals. New York. Wiley. 476 pp.
Forester, D.C. and A. Wisnieski. 1991. The significance
of airborne olfactory cues to the recognition of home
area by the Dart-poison frog Dendrobates pumilio.
Journal of Herpetology 25: 502–504.
Grafen, A. 1990. Do animals really recognize kin? Animal
Behaviour 39: 42–54.
Hamilton, W. D. 1964. The genetical evolution of social
behavior I and II. Journal of Theoretical Biology 7: 1–
52.
Meffe, G. K. and M. L. Crump. 1987. Possible growth and
reproductive benefits of cannibalism in the
mosquitofish. The American Naturalist 129: 203–212.
Nagai, Y., S. Nagai and T. Nishikawa. 1971. The nutritional
efficiency of cannibalism and an artificial feed for the
growth of tadpoles of Japanese toad (Bufo vulgaris
sp.). Agricultural Biology and Chemistry 35: 697–703.
Kin discrimination in cannibalistic tadpoles of the Green Poison Frog
Phyllomedusa
- 8(1), July 2009
50
Pfennig, D. W. 1997. Kinship and cannibalism. Bioscience
47: 667–675.
Pfennig, D. W. and J. P. Collins. 1993. Kinship affects
morphogenesis in cannibalistic salamanders. Nature
362: 836–838.
Pfennig, D. W. and W. A. Frankino. 1997. Kin-mediated
morphogenesis in facultatively cannibalistic tadpoles.
Evolution 51: 1993–1999.
Pfennig, D. W., M. L. G. Loeb and J. P. Collins. 1991.
Pathogens as a factor limiting the spread of canniba-
lism in tiger salamanders. Oecologia 88: 161–166.
Pfennig, D. W., H. K. Reeve and P. W. Sherman. 1993.
Kin recognition and cannibalism in spadefoot toad
tadpoles. Animal Behaviour 46: 97–94.
Pfennig, D. W., S. G. Ho and E. A. Hoffman. 1998.
Pathogen transmission as a selective force against
cannibalism. Animal Behaviour 55: 1255–1261.
Pfennig, D. W., J. P. Collins and R. E. Ziemba. 1999. A
test of alternative hypotheses for kin recognition in
cannibalistic tiger salamanders. Behavioral Ecology
10: 436–443.
Polis, G. 1981. The evolution and dynamic of intraspecific
predation. Annual Review of Ecology and Systematics
12: 125–151.
Sherman, P. W., H. K. Reeve and D. W. Pfennig. 1997.
Recognition Systems. Pp. 69–96 in J. R. Krebs and N.
B. Davies (eds.), Behavioural Ecology: an Evolutio-
nary Approach. 4th edition. London. Blackwell.
Sherratt, T. N., S. E. Ruff and S. C. Church. 1999. No
evidence for kin discrimination in cannibalistic tree-
hole mosquitoes (Diptera: Culicidae). Journal of Insect
Behavior 12: 23–128.
Silverstone, P. A. 1975. A revision of the poison-arrow
frogs of the genus Dendrobates Wagler. Bulletin of Los
Angeles County Museum of Natural History 21: 1–55.
Summers, K. 1989. Sexual selection and intra-female
competition in the green poison-dart frog, Dendrobates
auratus. Animal Behaviour 37: 797–805.
Summers, K. 1990. Paternal care and the cost of polygyny
in the green dart-poison frog. Behavioral Ecology and
Sociobiology 27: 307–313.
Summers, K. 1992. Mating strategies in two species of
dart-poison frogs: a comparative study. Animal
Behaviour 43: 907–919.
Summers, K. 1999. The effects of cannibalism on
Amazonian poison frog egg and tadpole deposition and
survivorship in Heliconia axil pools. Oecologia 119:
557–564.
Wakahara, M. 1997. Kin recognition among intact and
blinded, mixed-sibling larvae of a cannibalistic
salamander Hynobius retardatus. Zoological Science
14: 893–899.
Waldman, B. 1985. Olfactory basis of kin recognition in
toad tadpoles. Journal of Comparative Physiology A
156: 565–577.
Walls S. C. and A. R. Blaustein. 1995. Larval marbled
salamanders, Ambystoma opacum, eat their kin. Animal
Behaviour 50: 537–545.
Walls, S. C. and R. E. Roudebush. 1991. Reduced
aggression toward siblings as evidence of kin
recognition in cannibalistic salamanders. The American
Naturalist 138: 1027–1038.
Weygoldt, P. 1987. Evolution of parental care in dart
poison frogs (Amphibia: Anura: Dendrobatidae).
Zeitschrift für zoologische Systematik und
Evolutionsforschung 25: 51–67.
Wilbur, H. M. and R. D. Semlitsch. 1990. Ecological
consequences of tail injury in Rana tadpoles. Copeia
1990: 18–24.
Wildy, E. L., D. P. Chivers, J. M. Kiesecker, and A. R.
Blaustein. 1998. Cannibalism enhances growth in
larval long-toed salamanders (Ambystoma
macrodactylum). Journal of Herpetology 32: 286–289.
Zar, J. H. 1999. Biostatistical Analysis. 4th edition. New
Jersey. Prentice Hall. 929 pp.
Gray et al.
