Behavioral Ecology Vol. 10 No. 4: 436-443
© 1999 International Society for Behavioral Ecology
A test of alternative hypotheses for kin recognition in cannibalistic tiger salamanders
a Department of Biology, University of North Carolina, Chapel Hill, NC 27599, USA b Department of Biology, Arizona State University, Tempe, AZ 85287, USA
Address correspondence to D. W. Pfennig, Department of Biology, Coker Hall, CB#3280, University of North Carolina, Chapel Hill, NC 27599-3280, USA. E-mail: dpfennig{at}email.unc.edu
Received 22 March 1998; revised 11 August 1998; accepted 6 January 1999.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODS AND RESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The function of kin recognition is controversial. We investigated the adaptive significance of kin discrimination in cannibalistic tiger salamander larvae, Ambystoma tigrinum. Previous laboratory experiments show that cannibals preferentially consume less related individuals. We hypothesized that this example of kin recognition (1) is a laboratory artifact, (2) is a by-product of sibship-specific variation in escape responses, because cannibals from families with rapid responses may be more likely to cannibalize slowly escaping non-kin, (3) is an epiphenomenon of species recognition, (4) functions in disease avoidance, because kin may be more infectious than non-kin, or (5) is favored by kin selection. We evaluated these five hypotheses by using laboratory and field experiments to test specific predictions made by each hypothesis. We rejected hypotheses 1-4 above because (1) kin recognition was expressed in the wild, (2) escape responses did not reliably predict whether a cannibal would ingest kin or non-kin, (3) kin recognition was not most pronounced in populations where tiger salamanders co-occur with other species of salamanders, and (4) non-kin prey were more likely than kin to transmit pathogens to cannibals. However, we established that the necessary condition for kin selection, Hamilton's rule, was met. Thus, our results implicate kin selection as the overriding reason that cannibalistic tiger salamanders discriminate kin.
Key words: Ambystoma tigrinum, cannibalism, disease transmission, Hamilton's rule, kin discrimination, kin selection, salamanders.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODS AND RESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Kin recognition, the differential treatment of conspecifics varying in genetic relatedness, has been documented in multiple animal and plant taxa (reviewed in Fletcher and Michener, 1987
; Hepper,
1991; Pfennig and Sherman,
1995; Waldman,
1991). Although the catalyst to study kin recognition is
ultimately interest in its evolutionary cause
(Waldman, 1991), little is
known about the fitness consequences of kin discrimination (reviewed in
Blaustein et al., 1991;
Gamboa et al., 1991;
Sherman et al., 1997). This
gap in our knowledge has provoked a spirited debate regarding the evolutionary
role of kin recognition (Blaustein et al.,
1991; Grafen,
1990; Sherman et al.,
1997; Stuart,
1991) and has even led some to speculate that most organisms do
not really recognize kin at all (Grafen,
1990).
Both adaptive and nonadaptive hypotheses have been proposed to explain kin
recognition (Barnard, 1991;
Blaustein et al., 1987;
Carlin, 1989;
Grafen, 1990;
Pfennig, 1990;
Pfennig et al., 1993;
Waldman, 1991). In particular,
kin recognition has been purported to be (1) an epiphenomenon of some other
recognition system, (2) maintained by natural selection because it enhances
the direct component of a discriminator's inclusive fitness (i.e., the genes
contributed to the next generation by an individual via personal
reproduction), or (3) maintained by natural selection because it enhances the
indirect component of a discriminator's inclusive fitness (i.e., the genes
contributed to the next generation by an individual indirectly by helping
nondescendant kin; Brown,
1987). These three hypotheses differ from one another in the
fitness benefits derived by discriminating individuals. The epiphenomenal
hypothesis posits that discriminators do not benefit by their actions, whereas
the two selective hypotheses differ in the type of benefit that discriminators
accrue.
Cannibalistic species are ideal for testing alternative hypotheses for kin
recognition because kin discrimination may be particularly well developed in
these species (Pfennig, 1997,
1999;
Pfennig and Collins, 1993;
Pfennig et al., 1993,
1994;
Sadler and Elgar, 1994;
Wade, 1980;
Walls and Roudenbush, 1991).
Indeed, parents refrain from preying on their own offspring but readily
cannibalize less related young in numerous cannibalistic species, and in some
species, even collateral, or nondescendant, kin are avoided (reviewed in
Pfennig, 1997; but see
Walls and Blaustein, 1995, and
references therein).
There are at least five hypotheses to explain why cannibals avoid preying on kin (Table 1). The first three hypotheses are epiphenomenal hypotheses, whereas the last two are selective hypotheses that differ in the type of benefit (direct or indirect) that a discriminating cannibal receives.
|
First, kin recognition might be an artifact of laboratory conditions
(Gamboa et al., 1991). In
laboratory tests (e.g., Pfennig et al.,
1994), cannibals are given a choice of eating equal numbers of kin
and non-kin. Kin recognition may be absent in more complex natural settings,
where environmental heterogeneity and more diverse prey choices might preclude
a cannibal's distinguishing kin from non-kin. Second, kin recognition might be
a by-product of sibship-specific variation in escape responses. Cannibals from
different sibships might vary in both the speed they attack prey and the speed
of reaction to attacks from other cannibals. Thus, cannibals from families
with rapid responses would be more likely to cannibalize slowly escaping
non-kin than rapidly escaping kin, giving the spurious impression that
cannibals prefer to eat non-kin. Third, kin recognition might be an
epiphenomenon of species recognition
(Grafen, 1990). In laboratory
choice tests (Pfennig et al.,
1994), cannibals are often raised with siblings only. Thus, they
might learn their species recognition cues or "template" from
siblings. If so, their avoidance of siblings might represent attempts to avoid
consuming conspecifics. Indeed, when given a choice of preying on conspecific
or heterospecific salamander larvae, tiger salamanders prefer to eat the
latter (Pfennig et al.,
1998).
Fourth, kin recognition might function in disease avoidance
(Pfennig et al., 1993). There
are numerous accounts of parasites being transmitted via cannibalism (reviewed
in Pfennig et al., 1998;
Polis, 1981). Parasites are
often strongly host specific, apparently because of coevolution between
parasites and hosts (reviewed in Freeland,
1983; Møller et al.,
1993). Thus, genetically similar organisms may be especially
likely to exchange parasites. For example, cannibalistic tiger salamander
larvae are more likely to acquire pathogens from conspecifics than from
heterospecifics (Pfennig et al.,
1998). Similarly, close relatives may be more likely than
nonrelatives to exchange parasites, owing to greater genetic similarity among
close relatives and selection for host specificity and resistance to host
immune defenses among pathogens. Indeed, there is evidence from bumblebees and
humans that certain parasites are more highly transmissible among kin than
among non-kin (Black, 1994;
Shykoff and Schmid-Hempel,
1991). Thus, ingesting close relatives may be costly to cannibals
because kin may be more infectious than non-kin.
Fifth, kin recognition might be maintained by kin selection
(Hamilton, 1964). In
particular, a cannibal that recognizes and avoids preying on kin may act in
its own genetic self-interest by propagating genes shared with kin, including
genes for kin recognition. In more precise terms, kin recognition will be
selectively favored whenever Hamilton's rule is satisfied
(Hamilton, 1964); i.e.,
whenever rb - c > 0, where r is the coefficient of relatedness between
discriminator and its potential prey, c is the cost of the act in terms of
future offspring production that the discriminator loses by not eating the
prey, and b is the benefit of the act in terms of the extra offspring that
noncannibalized prey gain. Thus, kin recognition may be favored because of the
indirect fitness benefits that a discriminatory cannibal accrues.
We evaluated the above five alternative hypotheses for kin recognition in
cannibalistic tiger salamanders (Ambystoma tigrinum), a species that
often occurs in nature as a typical morph that feeds mostly on invertebrates
and occasionally on other salamanders and as a physically distinctive cannibal
morph that preys mostly on conspecifics
(Collins et al., 1993;
Powers, 1907). Cannibal morphs
are produced when larvae are crowded with other salamanders
(Collins and Cheek, 1983;
Hoffman and Pfennig, 1999). In
laboratory choice tests, cannibalistic larvae use sibship-specific olfactory
signals to feed voraciously on nonrelatives but avoid eating close kin
(Pfennig et al., 1994). They
also are significantly more likely to express the cannibal phenotype in mixed
sibship groups than in pure sibship groups
(Pfennig and Collins, 1993).
To determine which of the above five hypotheses explains these well-developed
kin recognition abilities, we tested specific predictions made by each
hypothesis (Table 1).
|
METHODS AND RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODS AND RESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Experimental animals
We studied A. tigrinum larvae from the White Mountains of Arizona and West Lafayette, Indiana, USA (for locations and descriptions of White Mountains ponds, see Pfennig et al., 1994
). For the White Mountains populations, we used larvae from 10
sibships whose parents or grandparents were captured from 6 ponds. Each pair
of adults was kept in a 120-1 aquarium until the female laid eggs. For the
West Lafayette population, we used larvae from 10 sibships whose parents were
from a single pond. Adults were captured at a drift fence as they approached
the pond to breed. We assigned males and females to pairs and placed them
inside small, mesh cages that were partially submerged in the breeding pond.
After the female oviposited, the eggs were removed. Once the eggs from both
populations had hatched, we randomly chose 10-20 groups of 15 larvae from each
sibship. Groups were placed into separate aquaria (22 1) with 16 1 of
dechlorinated tap water. All larvae were reared under identical conditions:
22°-25°C water temperature and 14 h:10 h photoperiod. During rearing,
water was changed weekly and animals were fed ad libitum live brine shrimp
(Artemia sp.) daily. At 7 weeks after hatching we scored larvae as
being typical or cannibal morphotypes (the latter are characterized as having
an enlarged vomerine ridge and elongate teeth;
Pedersen, 1991).
Experiment 1: Is kin recognition an artifact of the laboratory?
Methods
To determine if tiger salamanders recognize kin in a natural setting, we
tested kin discrimination abilities of cannibals from three Arizona sibships
when they were 7 weeks old (equivalent to the age of test cannibals in
previous laboratory studies; Pfennig et
al., 1994). The experiment was conducted in Dude Lake, a natural
pond on the Mogollon Plateau of Arizona. This pond contains many cannibals,
and it is free of salamander disease epidemics
(Loeb et al., 1994;
Pfennig et al., 1991a). To
start the experiment, we positioned 20 cylindrical mesh cages (0.75 m wide
x 1.3 m deep) in 0.3 m-deep water. The cages were made of mesh small
enough (2 mm) to retain salamander larvae, but large enough to admit naturally
occurring prey (e.g., plankton and aquatic insects).
In each of 18 enclosures, we placed 25 larvae: 1 cannibal and 6 typical
morph larvae from each of 4 different sibships (24 typicals total). Six of
these typical morph larvae were the cannibal's siblings; the remaining 18 were
nonsiblings. Typical morph larvae within each enclosure were matched for size;
they were about half the snout-vent length (SVL) of the test cannibal. Typical
morph larvae had been reared apart from their cannibal siblings since they
were 
2 weeks old; different sibships were completely unfamiliar with each
other. In each of the two remaining enclosures, we placed six typical-morph
larvae from each of the four sibships. These served as controls to determine
if larvae from different sibships differed in mortality rate.
To keep track of kinship identities, we used a 26-gauge hypodermic needle to inject into the dorsal tail membrane of each larva a mixture of fluorescent pigment and oil (one part mineral oil to one part petroleum jelly). Once injected, the mark formed a thin strip (approximately 1 mm wide x 20 mm long) of red, yellow, pink, or orange fluorescent pigment. Within each enclosure, animals from different sibships were injected with different colors. These marks were visible under ordinary light until after metamorphosis and did not affect larval mortality. To control for any effects of different colors on a cannibal's prey preferences, the test cannibal's siblings were represented by each color type in different enclosures.
The experiment began on 10 August 1994. An observer (unaware of the sibship
identities of the stimulus animals) checked each enclosure approximately
weekly and noted the tail marks of surviving stimulus animals. We inferred
that cannibalism had occurred if a larva was missing. Censuses were conducted
on 20 August, 1, 10, 17, 23 September, and 2 October (by which time all
cannibals had metamorphosed). Each cannibal's age and size (SVL) at
metamorphosis were also recorded. For each cannibal, the response variable was
the percentage of prey that were siblings. We predicted that if cannibals did
not discriminate kin (the null hypothesis), then 25% of prey would be
siblings. Proportional data were arcsine-square-root transformed to meet
parametric assumptions of normality. An initial analysis of sibship effects
indicated no significant differences among sibships in discriminatory ability
(F2.17 = 1.06, p =.371). We therefore treated
different sibships as replicates and used a one-sample t test to
compare the percentage of prey that were siblings with 25%. We used a
one-tailed test because a previously published report
(Pfennig et al., 1994)
indicated that tiger salamander larvae from the same population avoid eating
their kin in the laboratory.
Results
Any differences in mortality between kin and non-kin in treatment
enclosures could be ascribed to prey preferences of cannibals, and not to
variation among families in other sources of mortality because all typical
morph larvae survived in control enclosures. As shown in
Table 2, the mean percentage of
prey that were siblings in treatment enclosures was 20%, which was
significantly less than 25%, the value expected if cannibals had eaten
siblings and nonsiblings at random (t = -2.09, df = 17, p
=.026; one-tailed, one sample t test on arcsine-square-root
transformed data). Thus, kin recognition is expressed in the wild,
demonstrating that avoidance of kin cannibalism is not a by-product of
laboratory conditions.
|
Experiment 2: Is kin recognition an artifact of sibship-specific
variation in escape response?
Methods
We first determined if families varied in escape responses by subjecting
similarly sized typical morph larvae from different families to a simulated
attack by a cannibal (see Pfennig et al.,
1998). We used 95 larvae from 5 Arizona sibships that were housed
individually in 22-1 tanks filled with 16 1 of dechlorinated water. An
observer, who was unaware of a larva's sibship, touched the distal tail of
each larva with a plastic rod (460 mm long x 10 mm diameter) and
assigned each animal to an ordinal score depending on its response to being
tapped. If the salamander showed no response, we gave it a score of 0, if the
salamander crawled away, we gave it a score of 1, and if the salamander swam
away, we gave it a score of 2. We determined if families differed in these
responses by using a Kruskal-Wallis nonparametric one-way ANOVA test. We then
ranked the families according to their mean escape response scores, from
lowest (= slowest) to highest (= fastest).
Next, we determined if family-specific variation in escape response
predicts whether a cannibal would ingest kin or non-kin. We evaluated kin
discrimination abilities of 63 cannibals from the same 5 families that were
used to calculate escape response scores. Cannibals were given a choice of
eating a sibling larva or a similarly sized nonsibling larva, the latter being
from one of the other four families. We predicted that if kin recognition were
an artifact of sibship-specific variation in escape response, then the
relative escape response scores of the two families involved should reliably
predict whether the cannibal would eat kin or non-kin. For instance, if the
cannibal's family were slower than the non-kin's family in escape response,
then the escape-response hypothesis would predict that the cannibal should eat
(slower) kin instead of (faster) non-kin. We asked how many times this
prediction was met for the 20 combinations of cannibal-non-kin families. We
tested each cannibal when it was satiated (i.e., when cannibals had been fed
another salamander 1-2 days before the test took place) and again when it was
hungry (i.e., when cannibals had not been fed another salamander 7-8 days
before the test took place). We controlled for hunger because of a previous
report (Pfennig et al., 1993)
that cannibalistic spade-foot toad tadpoles are less likely to avoid eating
siblings when they are hungry than when they have just eaten.
To start an experiment, we put 16 1 of dechlorinated tap water into a 22-1
aquarium and introduced one cannibal morph and two "stimulus"
animals (both typical morph larvae) matched for size. One stimulus animal was
the cannibal's sibling, and the other was a nonsibling. Stimulus larvae were
about half the cannibal's SVL, and stimulus animals had been reared apart from
cannibals since they were 2 weeks old; different sibships were completely
unfamiliar with each other.
To keep track of kinship identities, we cut a small hole (2-3 mm) in either the dorsal or ventral half of each stimulus animal's fin. To control for effects of these marks, in half the aquaria the test cannibal's sibling was marked dorsally and the nonsibling ventrally, and in the other half the test cannibal's sibling was marked ventrally and the nonsibling dorsally. These marks did not affect larval mortality. An observer who was unaware of the sibship identities of the stimulus animals checked each aquarium at least once every hour between 0800 h and 2000 h and noted when cannibalism had occurred (when a tank mate was consumed) and the tail mark of the surviving stimulus animal. Throughout the experiment, larvae were fed the standard ration of live Artemia daily. Test and stimulus animals were used only once. The response variable was the percentage of prey each cannibal ingested that were siblings. These proportions were arcsine-square-root transformed to meet parametric assumptions of normality.
Results
Although larvae from different families differed significantly in escape
responses (H = 13.72, df = 4, p =.008; Kruskal-Wallis test),
relative response scores did not predict reliably whether a satiated cannibal
would ingest kin or non-kin. In 20 trials, the slower prey was eaten 11 times
and the faster prey was eaten 9 times (
2 = 0.2, df = 1,
p =.66). However, when the cannibal was hungry, escape responses
correctly predicted whether kin or non-kin were eaten: in all 20 trials using
hungry cannibals, the cannibal ate the slower prey. Thus, the escape-response
hypothesis applies in the special case where cannibals are hungry and,
presumably, not discriminating kin. However, because the escape-response
hypothesis cannot explain avoidance of kin cannibalism when cannibals are
satiated, we reject this hypothesis as a general explanation for kin
recognition.
Experiment 3: Is kin recognition an epiphenomenon of species
recognition?
Methods
The species-recognition hypothesis predicts that cannibals from populations
that co-occur with other species of salamanders should have more refined kin
discriminatory abilities than cannibals from populations that do not co-occur
with other species of salamanders. We tested this prediction by contrasting
kin discriminatory abilities of 63 cannibals from 5 Arizona sibships with that
for 50 cannibals from 10 Indiana sibships. Cannibals from Indiana occur
sympatrically with other species of salamanders, including a congener, A.
texanum. In contrast, A. tigrinum is the sole salamander species
found in Arizona.
We tested the discrimination abilities of larvae 7-9 weeks after they had
hatched using the testing procedures outlined above in experiment 2. However,
we controlled for inter-regional variation in kin discriminatory ability due
to differing propensities to produce cannibals. This control was important
because, for Arizona larvae, the greater the probability that a larva from a
given sibship would develop into a cannibal morph, the more likely the members
of that sibship are to discriminate kin
(Pfennig et al., 1994). Thus,
it might be contended that any between-region variation in kin discriminatory
ability may be due to differing propensities to produce cannibals and not to
differing exposures to heterospecifics. We therefore contrasted kin
discriminatory abilities for sibships across the two regions that had similar
propensities to produce cannibals. The propensity of each sibship to produce
cannibal morphs was determined by calculating the proportion of 10-20 separate
aquaria containing each sibship that produced a cannibal morph (e.g., see
Pfennig and Collins,
1993).
Results
Of 50 cannibals tested from the population that occurs sympatrically with
other species of salamanders (i.e., Indiana population), 26 (52%) ate kin, but
of 63 cannibals tested from the allopatric population (i.e., Arizona
population), only 21 (33%) ate kin. However, these inter-regional differences
in kin discrimination disappeared once we controlled for inter-regional
variation in propensity to produce cannibal morphs: when we restricted our
comparison to those Arizona sibships that did not differ from Indiana sibships
in propensity to produce cannibal morphs (i.e., sibships 1-5 in
Figure 1), cannibals from the
two regions did not differ in mean discrimination abilities (2
= 2.49, df = 1, p =.11). Therefore, kin recognition was not more
pronounced in populations where tiger salamanders are sympatric with other
species of salamanders, implying that kin recognition is not an epiphenomenon
of species recognition.
|
Experiment 4: Does kin recognition function in disease
avoidance?
Methods
The disease-avoidance hypothesis may be especially applicable to tiger
salamanders because this species is often afflicted with deadly disease
epidemics (Worthylake and Hovingh,
1989; Pfennig et al.,
1991a; Jancovich et al.,
1997; Pfennig et al.,
1998). Moreover, cannibalism is a mode of disease transmission in
this system (Pfennig et al.,
1991a; Pfennig et al.,
1998). Although the precise causes of these epidemics are not
known, disease may be caused by two species of bacteria
(Acinetobacter sp.: Worthylake
and Hovingh, 1989; Clostridium sp.:
Pfennig et al., 1991a) and a
virus (Ambystoma tigrinum virus:
Jancovich et al., 1997).
To determine if kin recognition functions in disease avoidance, we first
examined whether families differ in susceptibility to disease. We randomly
selected 16 equal-sized, 7-week-old typical morph larvae from each of 6
Arizona sibships that had been reared under similar conditions since birth and
placed them individually into a container filled with 3.6 1 of dechlorinated
tap water. We then exposed eight larvae from each sibship (treatment larvae)
to 0.4 1 of water from a diseased pond, and eight larvae (control larvae) to
0.4 1 of autoclaved water from the same pond. We fed each treatment animal 2 g
of liver from a diseased animal (livers of diseased animals have numerous
lesions containing pathogenic bacteria;
Pfennig et al., 1991a).
Control larvae were each fed 2 g of liver from a healthy animal. These control
larvae were used to determine if families differed intrinsically in their
survival, whereas treatment larvae were used to determine if families differed
in susceptibility to disease. Because the pathogens are often highly virulent,
the response variable was number of days after the start of the experiment
when treatment animals died. We used a one-way ANOVA to determine if
individuals from different families varied significantly in time of death.
We then asked if cannibals from diseased and nondiseased populations differ in kin discrimination abilities. The disease avoidance hypothesis predicts that cannibals from diseased populations should avoid eating kin, whereas those from non-diseased populations should eat kin and non-kin indiscriminately. We tested this prediction by contrasting the kin discrimination abilities of cannibals from Arizona and Indiana. Disease epidemics are common in many parts of Arizona, including the White Mountains where our experimental subjects were collected (Collins JP, personal observation). In contrast, disease epidemics have not been reported in Indiana. We conducted this experiment simultaneously with experiment 3 by using 63 cannibals from 5 Arizona sibships and 50 cannibals from 10 Indiana sibships. The response variable was the percentages of prey each cannibal ingested that were siblings.
Next, we tested the critical prediction of the disease avoidance
hypothesis: that kin are more infectious than non-kin. We compared disease
transmission among kin and non-kin in 1993 using 32 similarly sized cannibals
from 4 Arizona sibships and again in 1994 using 30 similarly sized cannibals
from 3 additional Arizona sibships. Cannibals were housed individually in
aquaria and reared under identical conditions (22°-25°C water
temperature and 14 h:10 h photoperiod; they were fed live brine shrimp daily
ad libitum). We randomly assigned cannibals to two different prey treatment
groups: each cannibal was either fed a single diseased sibling (1993:
n = 17; 1994: n = 16) or a single diseased nonrelative
(1993: n = 15; 1994: n = 14). Prey were similarly sized
typical morph larvae that were about half the size of the cannibals. We
created diseased prey by housing typical morph larvae inside aquaria with
diseased field-caught larvae for 2 h (larvae exposed longer invariably died of
disease). In 1993 larvae were exposed to diseased field-caught larvae with
symptoms of Ambystoma tigrinum virus
(Jancovich et al., 1997),
whereas in 1994 larvae were exposed to diseased field-caught larvae with
symptoms of Clostridium bacterial infection
(Pfennig et al., 1991a). We
anticipated that many cannibals would die because ingestion of a single
diseased larva is sufficient to cause mortality due to disease
(Pfennig et al., 1991a).
Therefore, the response variable was number of cannibals that survived to
metamorphosis.
Finally, we tested the prediction that cannibals should prefer nondiseased over diseased prey. We placed 20 satiated cannibals from Arizona individually in 22-1 aquaria filled with 16 1 of dechlorinated water. Into each tank, we placed two smaller, non-kin, typical morph larvae, one of which was healthy and the other of which was diseased (diseased animals display hemorrhagic septicemia, and they tend to float lethargically on the water surface). An observer recorded which animal (diseased or healthy) was consumed first. We used a chi-square test to compare the number of each type of prey consumed to the number expected (10) if cannibalism were random with respect to prey health.
Results
In the test to determine whether families differed in susceptibility to
disease, all treatment larvae died. However, larvae from different families
differed significantly in time of death (F5,44 = 6.23,
p =.0002; Table 3).
There were even highly significant sibship effects on the time of death among
the three LC sibships that were cousins (F2,22 = 16.32,
p =.0001; Table 3).
That these family-specific differences were likely due to disease and not to
intrinsic differences in mortality was suggested by the finding that all
control larvae survived. Thus, families differed in susceptibility to disease,
and this variation occurred among larvae from the same pond.
|
Given that the above critical assumption of the disease avoidance hypothesis was true, we next tested this hypothesis in three ways. First, we used a comparative approach to determine if cannibals from diseased and nondiseased populations differed in kin discrimination abilities. We found that once we had controlled for inter-regional variation in propensity to produce cannibal morphs, cannibals from diseased and nondiseased regions did not differ in discrimination abilities (Figure 1, see also experiment 3 results). Therefore, contrary to the prediction of the disease avoidance hypothesis, kin recognition was not more pronounced in diseased populations than in nondiseased populations.
Second, we asked if kin are more infectious to cannibals than are non-kin.
We found that in 1993, 5 of 15 (33%) cannibals that ate diseased non-kin died,
whereas none of the 17 cannibals that ate diseased kin died (2
= 6.716, df = 1, p =.009). In 1994, the results were similar,
although not statistically significant: 11 of 14 (79%) cannibals that ate
diseased non-kin died, whereas 9 of 16 (56%) cannibals that ate diseased kin
died (2 = 1.674, df = 1, p =.20). When both years
were combined, cannibals of non-kin died at a significantly higher rate than
did cannibals of kin (Table 4).
Therefore, contrary to the prediction of the disease avoidance hypothesis,
diseased non-kin were more infectious than were diseased kin.
|
Finally, we predicted that if the disease avoidance hypothesis were
correct, cannibals should avoid diseased prey. However, when 20 satiated
cannibals from the diseased population were housed individually with two
smaller larvae, 1 diseased and 1 healthy, 13 cannibals (65%) ate the diseased
prey (presumably because they were easier to catch), and 7 cannibals (35%) ate
the healthy prey (2 = 1.98, df = 1, p =.16). Thus,
contrary to the prediction of the disease avoidance hypothesis, cannibals did
not avoid diseased prey.
Experiment 5: Is kin recognition kin selected?
Methods
To answer the question of whether kin recognition is kin selected, we asked
if avoidance of kin cannibalism satisfies Hamilton's rule
(Hamilton, 1964). Hamilton's
rule predicts that avoidance of sibling cannibalism will be favored by natural
selection if c/b <
, where
is the coefficient of
relatedness between full siblings. To estimate cost (c) and benefit
(b) of kin discrimination, we used various measures of direct and
indirect fitness to compare cannibals in experiment 1 that ate <25% kin
(discriminators, n = 13) with those cannibals that ate 
25% kin
(nondiscriminators, n = 5), where 25% kin consumption was the value
expected if cannibalism were random with respect to kinship (see
Table 2). Discriminators and
nondiscriminators were from the same sibships, they were of the same age and
initial sizes (mean±SD SVL of discriminators = 38.3±3.9 mm;
mean±SD SVL of nondiscriminators = 38.7±1.5; p =.66;
two-tailed Mann-Whitney test), and they had similar growth rates
(Figure 2). We predicted that
if discriminators fared significantly better than nondiscriminators in terms
of benefit, but if the two types of cannibals did not suffer different costs,
then Hamilton's rule would be true (i.e., c/b <
), implying that kin
recognition is kin selected.
|
To estimate benefit of kin discrimination, we compared the number of siblings that survived to metamorphosis for discriminators and nondiscriminators. Note that this benefit is not inevitably positive; for example, if all noncannibalized individuals were to die from starvation or disease before metamorphosis, then this benefit would be zero, and if noncannibalized siblings were more likely than nonsiblings to compete for the same foods, then this benefit could even be negative.
To estimate cost of discrimination, we compared survival, growth rate, and
age at metamorphosis for discriminators and nondiscriminators. Our rationale
for using these measures was that a cannibal that recognizes and avoids
preying on kin thereby provides benefits to its relatives, but the cannibal
may suffer the personal cost of diminished growth or survival by forgoing a
meal (Crump, 1992). Growth
rate and age at metamorphosis are both likely to be sensitive to reduced food
intake [such as what may be experienced by a discriminating cannibal (e.g.,
see Lannoo et al., 1989)], and
both correlate significantly with several components of fitness in amphibians,
such as adult survival (Pfennig et al.,
1991b) and age at first reproduction
(Semlitsch et al., 1988).
Results
As indicated in Figure 2,
the benefit of kin discrimination (b) was large: discriminators had
more than twice as many siblings survive to metamorphosis than did
nondiscriminators (p =.004; two-tailed t test). In contrast,
the cost of kin discrimination was small
(Figure 2). Discriminators and
nondiscriminators had equal survival (all survived to metamorphosis), growth
rate (SVL; discriminators: 8.87±3.92 mm, nondiscriminators:
9.75±5.05 mm; p =.78; two-tailed Mann-Whitney test), and age
at metamorphosis (discriminators: 137±8 days; nondiscriminators:
132±12 days; p =.45; two-tailed Mann-Whitney test). Therefore,
the small costs that cannibals incurred by discriminating kin were likely
outweighed by the important benefits that relatives received by not being
eaten. Thus, Hamilton's rule is likely to be true in our system (i.e.,
c/b < ),
suggesting that kin recognition is maintained by kin selection.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODS AND RESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Our study helps clarify the function of kin recognition in cannibalistic species. We found no evidence that kin recognition is epiphenomenal in tiger salamanders (experiments 1-3). In particular, we found that (1) kin recognition was expressed in a natural setting, demonstrating that avoidance of kin cannibalism is not an artifact of laboratory conditions, (2) escape responses did not reliably predict whether a satiated cannibal would ingest kin or non-kin, implying that kin recognition is not a by-product of sibship-specific variation in escape responses, and (3) kin recognition was not most pronounced in populations where tiger salamanders co-occur with other species of salamanders, ruling out species recognition. Moreover, it is unlikely that cannibals simply avoid eating salamanders they are reared with, as opposed to kin per se. Previous experiments show that cannibals are as effective at discriminating between first cousins and nonrelatives as they are discriminating between siblings and nonrelatives, despite never having been exposed to cousins (Pfennig et al., 1994
). Thus, although Grafen
(1990) has claimed that many
examples of kin discrimination are artifacts of some other recognition system,
such as species or group-member identification, kin recognition is not
epiphenomenal in cannibalistic tiger salamanders. Instead, kin recognition
appears to be maintained by natural selection because it enhances the
cannibal's inclusive fitness.
Given that kin recognition is maintained by natural selection, we asked
whether it enhances the direct or indirect component of a cannibal's inclusive
fitness. A possible direct benefit is disease avoidance
(Pfennig et al., 1993). This
hypothesis assumes that kin are more infectious than non-kin. However, we
found that unrelated prey were more likely than related prey to transmit
pathogens to cannibals (experiment 4). This finding, which was based on 2
years of data using two different types of pathogens, was surprising in light
of recent evidence that genetically similar individuals are more likely to
infect each other with pathogens than are genetically dissimilar individuals
(Black, 1994;
Pfennig et al., 1998;
Shykoff and Schmid-Hempel,
1991). It is unclear why non-kin would be more likely to transmit
pathogens to each other. Perhaps an individual's kin are less likely to carry
diseases to which the individual is susceptible but for which the individual
would not have previously developed immune responses. Regardless of why
non-kin are more infectious than kin, this finding weakens the disease
avoidance hypothesis.
Two other results from experiment 4 imply that kin recognition does not function in disease avoidance. First, when we controlled for different propensities to produce cannibal morphs, we found that cannibals from diseased populations were no more discriminating of kin than were cannibals from a nondiseased population (Figure 1). Second, when offered a choice of diseased prey and healthy prey, cannibals from diseased populations showed no preference, suggesting an absence of strong selection to avoid disease.
If kin recognition is neither an epiphenomenon nor a means for cannibals to
obtain direct inclusive fitness benefits, then kin selection furnishes a
compelling explanation for the evolutionary maintenance of kin recognition. By
Hamilton's rule (Hamilton,
1964), avoidance of sibling cannibalism will be favored when the
ratio of the cost incurred by the cannibal to the benefit accrued by the
recipient (i.e., c/b) is <
. We were able to assess the
relative values of c and b using data from an experimental
population in a natural setting. Such estimates of lifetime fitness often
suffice to test Hamilton's rule (e.g.,
Emlen and Wrege, 1989;
Grafen, 1984;
Reyer, 1984).
We inferred that Hamilton's rule is satisfied for avoidance of sibling
cannibalism (experiment 5). In particular, we found that discriminating
cannibals received substantial indirect fitness benefits by not eating kin,
but there was no evidence that these cannibals subsequently paid a cost by
avoiding kin cannibalism (Figure
2). Therefore, Hamilton's rule is likely to be satisfied in our
system, implying that kin recognition in cannibalistic tiger salamanders is
favored by kin selection. Thus, our study reinforces the view
(Blaustein and O'Hara, 1982;
Pfennig, 1999;
Pfennig and Collins, 1993;
Pfennig et al., 1993;
1994;
Sherman, 1981;
Wade, 1980;
Waldman, 1991;
Walls and Blaustein, 1995;
Walls and Roudenbush, 1991)
that kin selection is important for maintaining kin recognition in
cannibalistic individuals.
It might be contended that our estimate of the benefit of discrimination was inflated, since only one cannibal was in each of our field cages. In particular, if other unrelated cannibals were present, they might have eaten the focal cannibal's noncannibalized kin, thereby depressing the number of kin that ultimately survived to metamorphosis. This implies that in ponds where cannibals are present in high frequencies, Hamilton's rule may not be satisfied. Thus, the benefit of kin discrimination may differ in different environments.
It is also possible that the cost of kin discrimination varies in different environments. For instance, a discriminating cannibal may pay a substantial cost for discriminating kin in populations where disease epidemics occur because preferential cannibalism of non-kin appears to increase a cannibal's risk of acquiring deleterious pathogens (see experiment 4 results). Despite this cost, however, cannibals from diseased populations were no less discriminating than those from nondiseased populations (e.g., compare in Figure 1 the level of kin discrimination among cannibals from Arizona, where diseases are prevalent, with that among cannibals from Indiana, where diseases are apparently absent). Thus, the indirect inclusive fitness benefits of kin recognition must be substantial for this behavior to be maintained by natural selection even where diseases are prevalent.
In conclusion, despite the widespread view that kin selection is the primary agent maintaining kin recognition, relatively few studies have examined the fitness consequences of kin recognition. Although our results implicate kin selection as the overriding reason that cannibalistic tiger salamanders discriminate kin, it is important to consider whether kin recognition is an artifact or epiphenomenon of some other factor that operates instead of or in addition to kin selection. Only by considering alternative hypotheses are we likely to succeed in determining why organisms recognize their kin.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODS AND RESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
We thank Karin Pfennig, Kern Reeve, Paul Sherman, Ron Ydenberg, and two anonymous referees for helpful comments on the manuscript, Rick Howard and Howard Whiteman for collecting Indiana salamanders, Eric Hoffman, Larry Nienauber, Karin Pfennig, Jim Roth, and Kristine Ziemba for laboratory and field assistance, and the White Mountain Apache Indian tribe and the Arizona Game and Fish Department for collecting permits. This research was supported by grants from the U.S. National Science Foundation to D. P. and J. C.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODS AND RESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Barnard C, 1991. Kinship and social behaviour: the trouble with relatives. Trends Ecol Evol 6:310-312.
Black FL, 1994. An explanation of high death rates among New World peoples when in contact with Old World diseases.Perspect Biol Med 37:292-307.[Web of Science][Medline]
Blaustein AR, Bekoff M, Byers JA, Daniels TJ, 1991. Kin recognition in vertebrates: what do we really know about adaptive value?Anim Behav 41:1079-1083.
Blaustein AR, Bekoff M, Daniels TJ, 1987. Kin recognition in vertebrates (excluding primates): mechanisms, functions, and future research. In: Kin recognition in animals (Fletcher DJC, Michener CD, eds). New York: John Wiley;333 -357.
Blaustein AR, O'Hara RK, 1982. Kin recognition in Rana cascadae tadpoles: maternal and paternal effects. Anim Behav 30:1151-1157.
Brown JL, 1987. Helping and communal breeding in birds: ecology and evolution. Princeton, New Jersey: Princeton University Press.
Carlin NF, 1989. Discrimination between and within colonies of social insects: two null hypotheses. Neth J Zool 39:86-100.
Collins JP, Cheek JE, 1983. Effect of food and density on development of typical and cannibalistic salamander larvae in Ambystoma tigrinum nebulosum. Am Zool 23:77-84.
Collins JP, Zerba KE, Sredl MJ, 1993. Shaping intraspecific variation: development, ecology and the evolution of morphology and life history variation in tiger salamanders. Genetica 89:167-183.[Web of Science]
Crump ML, 1992. Cannibalism in amphibians. In:Cannibalism. Ecology and evolution among diverse taxa (Elgar MA, Crespi BJ, eds). Oxford: Oxford University Press;256 -276.
Emlen ST, Wrege PH, 1989. A test of alternative hypotheses for helping behavior in white-fronted bee-eaters of Kenya.Behav Ecol Sociobiol 25:303-319.[Web of Science]
Fletcher DJC, Michener CD (eds), 1987. Kin recognition in animals. New York: John Wiley.
Freeland WJ, 1983. Parasites and the coexistence of animal host species. Am Nat 121:223-236.[Web of Science]
Gamboa GJ, Reeve HK, Holmes WG, 1991. Conceptual issues and methodology in kin-recognition research: a critical discussion.Ethology 88:109-127.[Web of Science]
Grafen A, 1984. Natural selection, kin selection, and group selection. In: Behavioural ecology: an evolutionary approach, 2nd ed. (Krebs JR, Davies NB, eds). Sunderland, Massachusetts: Sinauer; 62-84.
Grafen A, 1990. Do animals really recognize kin?Anim Behav 39:42-54.[Web of Science]
Hamilton WD, 1964. The genetical evolution of social behaviour. I, II. J Theor Biol 7:1-52.[Web of Science][Medline]
Hepper PG, ed, 1991. Kin recognition. Cambridge: Cambridge University Press.
Hoffman EA, Pfennig DW, 1999. Proximate causes of cannibalistic polyphenism in larval tiger salamanders. Ecology 80:1076-1080.[Web of Science]
Jancovich JK, Davidson EW, Morado JF, Jacobs BL, Collins JP,1997 . Isolation of a lethal virus from the endangered tiger salamander Ambystoma tigrinum stebbinsi. Dis Aquat Org 31:161-167.
Lannoo MJ, Lowcock L, Bogart JP, 1989. Sibling cannibalism in non-cannibal morph Ambystoma tigrinum larvae and its correlation with high growth rates and early metamorphosis. Can J Zool 67:1911-1914.
Loeb MLG, Collins JP, Maret TJ, 1994. The role of prey in controlling expression of a trophic polymorphism in Ambystoma tigrinum nebulosum. Funct Ecol 8:151-158.
Møller AP, Dufva R, Allander K, 1993. Parasites and the evolution of host social behavior. Adv Study Behav 22:65-102.
Pedersen SC, 1991. Dental morphology of the cannibal morph in the tiger salamander, Ambystoma tigrinum.Amphib-Reptilia 12:1-14.
Pfennig DW, 1990. `Kin recognition' among spadefoot toad tadpoles: a side-effect of habitat selection? Evolution 44:785-798.[Web of Science]
Pfennig DW, 1997. Kinship and cannibalism.BioScience 47:667-675.[Web of Science]
Pfennig DW, 1999. Cannibalistic tadpoles that pose the
greatest threat to kin are most likely to discriminate kin. Proc R Soc
Lond B
266:57-61.
Pfennig DW, Collins JP, 1993. Kinship affects morphogenesis in cannibalistic salamanders. Nature 362:836-838.[Medline]
Pfennig DW, Ho SG, Hoffman EA, 1998. Pathogen transmission as a selective force against cannibalism. Anim Behav 55:1255-1261.[Web of Science][Medline]
Pfennig DW, Loeb MLG, Collins JP, 1991a. Pathogens as a factor limiting the spread of cannibalism in tiger salamanders.Oecologia 88:161-166.[Web of Science]
Pfennig DW, Mabry A, Orange D, 1991b. Environmental causes of correlations between age and size at metamorphosis in Scaphiopus multiplicatus. Ecology 72:2240-2248.[Web of Science]
Pfennig DW, Reeve HK, Sherman PW, 1993. Kin recognition and cannibalism in spadefoot toad tadpoles. Anim Behav 46:87-94.
Pfennig DW, Sherman PW, 1995. Kin recognition.Sci Am 272:98-103.[Web of Science][Medline]
Pfennig DW, Sherman PW, Collins JP, 1994. Kin
recognition and cannibalism in polyphenic salamanders. Behav
Ecol
5:225-232.
Polis GA, 1981. The evolution and dynamics of intraspecific predation. Annu Rev Ecol Syst 12:225-251.[Web of Science]
Powers JH, 1907. Morphological variation and its causes in Ambystoma tigrinum. Stud Univ Nebraska 7:197-274.
Reyer H-U, 1984. Investment and relatedness: a cost/benefit analysis of breeding and helping in the pied kingfisher.Anim Behav 32:1163-1178.
Sadler LM, Elgar MA, 1994. Cannibalism among amphibian larvae: a case of good taste. Trends Ecol Evol 9:5-6.
Semlitsch RD, Scott DE, Pechmann JHK, 1988. Time and size at metamorphosis related to adult fitness in Ambystoma talpoideum. Ecology 69:184-192.[Web of Science]
Sherman PW, 1981. Reproductive competition and infanticide in Belding's ground squirrels and other animals. In:Natural selection and social behavior (Alexander RD, Tinkle DW, eds). New York: Chiron Press; 311-331.
Sherman PW, Reeve HK, Pfennig DW, 1997. Recognition systems. In: Behavioural ecology: an evolutionary approach, 4th ed. (Krebs JR, Davies NB, eds). Oxford: Blackwell Scientific;69 -96.
Shykoff JA, Schmid-Hempel PS, 1991. Parasites and the
advantage of genetic variability within social insect colonies. Proc
Roy Soc Lond B
243:55-58.
Stuart RJ, 1991. Kin recognition as a functional concept. Anim Behav 41:1093-1094.
Wade MJ, 1980. An experimental study of kin selection.Evolution 34:844-855.[Web of Science]
Waldman B, 1991. Kin recognition in amphibians. In:Kin recognition (Hepper P, ed). Cambridge: Cambridge University Press; 162-219.
Walls SC, Blaustein AR, 1995. Larval marbled salamanders, Ambystoma opacum, eat their kin. Anim Behav 50:537-545.
Walls SC, Roudenbush RE, 1991. Reduced aggression toward siblings as evidence of kin recognition in cannibalistic salamanders.Am Nat 138:1027-1038.[Web of Science]
Worthylake KM, Hovingh P, 1989. Mass mortality of salamanders (Ambystoma tigrinum) by bacteria (Acinetobacter) in an oligotrophic seepage mountain lake. Great Basin Nat 49:364-372.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Segoli, T. Keasar, A. R. Harari, and A. Bouskila Limited kin discrimination abilities mediate tolerance toward relatives in polyembryonic parasitoid wasps Behav. Ecol., November 1, 2009; 20(6): 1262 - 1267. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. E. O'Connor and R. Shine Kin discrimination in the social lizard Egernia saxatilis (Scincidae) Behav. Ecol., March 1, 2006; 17(2): 206 - 211. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J. Stuart and J. M. Herbers Nest mate recognition in ants with complex colonies: within- and between-population variation Behav. Ecol., November 1, 2000; 11(6): 676 - 685. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||