Anim. Behav., 1997, 54, 493–502
Sibling recognition in the beaver: a field test for phenotype matching
LIXING SUN & DIETLAND MU} LLER-SCHWARZE
Department of Biology, College of Environmental Science and Forestry, State University of New York
(Received 6 June 1996; initial acceptance 31 August 1996;
final acceptance 13 November 1996; MS. number: A7624)
Abstract. The hypothesis of kin recognition by phenotype matching predicts that relatives can be
identified without previous contact, and/or that cues used for recognition can be learned indirectly from
a third but related individual. This hypothesis was tested in the field using 22 beaver, Castor canadensis,
families. Individually identifiable beavers were provided with a two-way choice between two experimental
scent mounds, one of which was scented with the anal gland secretion (AGS) from an unfamiliar
sibling of the test subjects, the other with AGS from an unfamiliar non-relative. Beavers showed less
strong territorial responses to AGS from their siblings than to that from non-relatives. The mates of the
test subjects, which were not related to, or familiar with, either of the AGS donors, also responded less
strongly to the AGS from their mates’ siblings than to that from other unfamiliar non-relatives. This
discrimination was not shown when castoreum samples were tested instead of AGS. Therefore, it was
concluded that (1) information about kinship in the beaver is coded in the AGS but not in the
castoreum, (2) the mechanism of phenotype matching is used in beaver sibling recognition, and (3) the
cue used in phenotype matching can be learned and used for recognition of related individuals by an
unrelated individual. ? 1997 The Association for the Study of Animal Behaviour
Hamilton (1964) proposes that an animal should
do a favour for its relatives according to the
decision rule b/c>1/r, where b and c are the benefit
to the recipient and cost to the donor associated
with the altruistic behaviour, respectively, and r is
the coefficient of relatedness between the benefit
donor and recipient. One assumption behind
Hamilton’s rule is that the degree of relatedness
(r) is recognizable, actively or passively. This
assumption has stimulated extensive testing for
kin recognition abilities in numerous species,
typically with a two-way choice setting in the
laboratory. Laboratory studies facilitate the control
of variables, but they may also confound
results with artefacts, two of which are highlighted
by Waldman (1987): subjects may be presented
a task that does not naturally exist, and the
ontogeny of kin recognition is often ignored.
Laboratory environments may also modify the
physiological status, and hence alter the response
pattern, of subjects. These constraints are particularly
true for mammals, which show a higher
degree of behavioural plasticity than any other
animal taxon. Field studies are therefore important
for consolidating and integrating laboratory
findings in a natural context.
The beaver, Castor canadensis, is among the few
mammalian species that are excellent models for
behavioural studies in the field (Mu¨ ller-Schwarze
& Houlihan 1991; Schulte 1993). It is a centralplace
forager (Jenkins 1980) with a small activity
core area consisting of one or several connected
ponds, which are also its territory. These traits
facilitate close-range observation and accurate
data acquisition. Beavers live in family units,
consisting of an adult pair and their offspring. The
two breeding adults give birth to 3–4 kits on
average in early summer every year. Two-year-old
male and female offspring disperse from their
natal families (reviewed in Jenkins & Busher
1979). Dispersal is assumed to occur along
streams (Townsend 1953; Leege 1968). Because
breeding adults can live as long as 9 years in the
same place (Svendsen 1989) and dispersers appear
to resettle in a suitable habitat that is close to their
natal families (L. Sun unpublished data), dispersers
of different years from the same family are
likely to meet outside their natal colony site
without knowing each other. Since altruism and
avoidance of extreme inbreeding can increase
fitness (Hamilton 1964; Bensch et al. 1994), this
benefit calls for mutual recognition between
related individuals without previous contact.
Social interactions between beavers are largely
regulated by chemical signals. At some locations
along the shoreline of beaver ponds, beavers regularly
place piles of mud or grass and then mark
them with secretions to advertise their territory
(Mu¨ ller-Schwarze & Heckman 1980; Svendsen
1980; Mu¨ ller-Schwarze & Houlihan 1991). Members
of different families have little interaction
except for late spring, when dispersal starts. During
this time, residents mark vigorously to advertise
their territoriality, and tens of fresh scent
marks can be found at one time. Marking activity
tapers off later, and only a few marks can be
found in late summer and autumn (Hodgdon
1978; Houlihan 1989). New territory owners,
however, frequently mark their territories independent
of seasons (L. Sun, unpublished data).
Therefore, the presence of scent marks signal the
ownership of the area.
The beaver has two well-defined pheromone
sources, anal gland secretion (AGS) and castoreum,
both located at the cloaca area. The anal
gland is sebaceous with wax esters and fatty acids
being major constituents. The gland is sexually
dimorphic and is not influenced by age (Sun
1996). The castor sac where castoreum is stored,
however, lacks secretory epithelium (Walro &
Svendsen 1981). Evidence shows that castoreum is
composed of compounds that are probably
dietary derivatives (Mu¨ ller-Schwarze 1992) and
seems not to be sexually dimorphic. Beavers
actively deposit castoreum while marking, but
how the AGS is deposited needs to be clarified. A
strange conspecific’s AGS or castoreum can elicit
territorial responses (Hodgdon 1978; Welsh &
Mu¨ ller-Schwarze 1989) but secretions from a
member of the same family have little effect in
eliciting behavioural responses (B. Schulte, personal
communication). Information about sex or
individuality are suspected to be contained in
beaver secretions (Hodgdon 1978; Welsh &
Mu¨ ller-Schwarze 1989).
Kin can be recognized by the following mechanisms:
(1) recognition by spatial distribution, (2)
recognition by social learning, (3) recognition by
phenotype matching, and (4) recognition by recognition
alleles (reviewed in Halpin 1991). Mechanism
1 is a passive process and not applicable to
our scenario. Mechanism 4 is of theoretical interest
but may not exist in the real world (Blaustein
1983; Holmes & Sherman 1983; Waldman 1987).
Mechanism 2 requires previous contact, which is
unlikely to be involved in recognizing unfamiliar,
earlier- or later-born relatives. In mechanism 3, an
animal recognizes its relatives by comparing the
similarity between a cue it has learned earlier
(template) and the label(s) the relatives carry. Kin
recognition through this process does not require
previous encounter, and thus phenotype matching
is most likely to be used by beavers to recognize
unfamiliar relatives outside their natal families. If
beaver secretions are used in kin recognition by
the mechanism of phenotype matching, we predicted
that (1) beavers show reduced territorial
responses to secretions from relatives even without
previous contact, and (2) cues used for discrimination
can be learned from a third individual
that is related to the test subject (Holmes &
Sherman 1983; Blaustein et al. 1987). If the second
prediction is true, we expected that the cues used
for kin discrimination between relatives can also
be learned by unrelated individuals. To test the
two predictions, we conducted a field experiment
using the method of two-sample choice test.
We trapped 116 beavers using Hancock live
traps baited with aspen, Populus tremuloides,
twigs at Allegany State Park, New York, from
1992 to 1995. We weighed and anaesthetized
captured beavers with a xylazine and ketamine
mixture injection (0.67 and 6.7 mg/kg body
weight, respectively). After immobilization, we
sexed beavers by the colour and viscosity of AGS
(Schulte et al. 1995) and cross-validated the results
by palpating for the presence or absence of the os
penis (Osborn 1955). We assigned beavers to age
classes (kit, 1 year old, 2 year old and adult) based
on body weight and size (Schulte 1993). Both ears
of each individual were tagged with a unique
colour combination of anodized aluminium ear
tags (National Band & Tag Co., Kentucky). Thus,
all beavers in the 22 families (average family
size=3.5; range=2–6) used in our study were
individually identifiable. Because trapping had
been carried out every year since 1984, and the life
history of most beavers had been monitored from
494 Animal Behaviour, 54, 3
a few months after they were born, the exact
genealogical relationships within the 22 famiies
were known. We milked secretion samples from
the left and right anal glands and castor sacs
separately into three 10-ml vials and allowed
2 hours for beavers to fully recover from the
anaesthetics before release. We dissolved each
secretion sample with methylene chloride (1:5 by
volume) and stored it in a "20)C freezer until
use. For the field bioassay, we diluted the solution
with methylene chloride (1:4 by volume) and then
used 0.25 ml for each application. Pilot studies
showed that this concentration was well above the
response threshold of the beaver.
To investigate discriminatory abilities of the
beaver, we designed a two-sample choice test and
measured territorial responses of resident beavers
as the dependent variables. We constructed two
experimental scent mounds and treated them with
beaver secretions to mimic the presence of intruders
(Fig. 1). We chose 30 cm as the distance
between the two scent mounds to ensure that once
a beaver responded to one of them, it would also
respond to the other; hence, between-treatment
effect could be compared. We used a size 18 cork
in each experimental scent mound to hold scent
materials and to control the evaporation surface
area (12 cm2).
Using the above experimental setting, we compared
beaver responses to scents (AGS or castoreum)
from unfamiliar siblings with responses
to those from unfamiliar non-siblings. Scent
donors were siblings with at least 2 years’ age
difference, so they had had no previous interactions
because the older siblings dispersed before
the younger ones were born (Fig. 2). We played
back the scents from older siblings to the younger
ones (at their natal sites) and the scents from the
younger siblings to the older ones (at the older
siblings’ new sites). In both situations, we treated
one experimental scent mound with the scent from
an unfamiliar sibling (treatment) and the other
scent mound with the scent from an unfamiliar
non-sibling collected from a remote site (>6 km;
control). We defined a remote site as that beyond
the average disperal distance (6.21 km: L. Sun,
In the second test, we used the same scent
donors and settings as the above, except that the
recipients were the older siblings’ mates, which
were not genetically related to, or had had any
previous contact with, either of the scent donors
(Fig. 2). This experiment was to test whether
beavers can use cues that they experienced from
their mates to discriminate between scents from
unfamiliar, younger siblings of their mates and
scents from unfamiliar non-relatives.
We used 10 families from 1994 and 12 families
from 1995 for bioassays in the spring and summer,
when marking activities were relatively high. We
reapplied the same pair of secretion samples but
with randomized assignments to the two experimental
scent mounds to bioassay at each family
on 6 consecutive days. Each beaver family was
used only once per year. Because AGS profiles are
sexually dimorphic, secretion samples used in the
experiments were from adult males (§2 years) to
avoid possible biases.
Recording and Data Analysis
We set up two pairs of experimental scent
mounds, one for AGS and the other for castoreum,
at two different locations on the bank of
a beaver pond about 30 min before beavers began
their late afternoon activity (at 1700–1800 hours).
We used a Tandy 102 laptop computer loaded
with the program Observer 2.0 (Noldus Information
Technology, The Netherlands) to record each
response pattern and duration. In a few cases of
multiple responses, where two or more beavers
responded sequentially to experimental scent
mounds, we included only the response of the first
beaver in our analyses because physical damage
to the scent mound (pawed, flattened or obliterated)
might cause some carry-over biases in the
following responses. Observation ended at 2100
hours, when fading daylight prevented individual
A complete beaver response to experimental
scent mounds can be sequentially separated into
sniffing (normally within 30 cm of the scent
mound), straddling (standing on the mound on
hind feet), pawing and, occasionally, overmarking
(putting a pile of mud either at the side or on top
of the original scent mound and then marking it
with castoreum and probably with AGS, too).
Weaker responses, however, consist of only the
earlier parts of the sequence. We thus used
response completeness (the number of different
behavioural patterns in a response) to measure
response intensity. Because the latter three patterns
were performed less frequently and involved
direct contact with the experimental scent
mounds, they were pooled and collectively named
‘physical response’. Normally, we included only
one individual from each family in our analyses.
Because responses to treated and control samples
were paired and the variables of interest were the
difference in response of the two scent mounds,
rather than the absolute time responding to
each of the scent mounds, occasional repeated
measurements should not mask beavers’ discriminatory
abilities. Since the response time was not
normally distributed, we used the Wilcoxon
signed-ranks test to examine the treatment effect.
The sign test was used to determine the significance
of the difference in beavers’ response completeness.
For each treatment, we compared
differences between the two sexes for sniffing,
physical and total responses using the Wilcoxon
two-sample test. We later pooled male and female
responses in our analyses because no significant
difference was detected. Since our hypothesis predicts
that beavers show reduced territorial behaviour
due to relationship or familiarity, all tests
were one-tailed with a significance level of 0.05,
Beavers (dispersers or the siblings of dispersers)
spent significantly less time responding to siblings’
AGS than to non-relatives’ AGS (Wilcoxon
signed-ranks test: T=66, N=22, P<0.020). This
was due to a shorter physical response to (T=65,
P<0.030), but not sniffing (T=102.5, P>0.100),
siblings’ AGS (Fig. 3). We also found a lower
degree of response completeness when beavers
responded to siblings’ AGS than to non-relatives’
AGS (sign test: C=5, N=22, P<0.030; Fig. 4).
Beavers showed no significant discrimination
when castoreum samples were tested, however (for
sniff: T=23, N=12, P>0.100; for physical
response: T=27, P>0.100; for total response:
T=30.5, P>0.250; Fig. 3; sign test: C=5, N=12,
P>0.300 for response completeness; Fig. 4).
The mates of dispersers also showed a lower
total response to the AGS samples from dispersers’
siblings than to those from dispersers’ nonrelatives
(T=88, N=25, P<0.030). They spent
more time sniffing (T=94.5, P<0.040) and physically
responding (T=99, P<0.050) to the AGS
from dispersers’ non-relatives than to dispersers’
siblings (Fig. 5), although the difference in
response completeness was marginally nonsignificant
(sign test: C=9, N=25, P=0.050; Fig.
6). Again, when castoreum samples were presented,
beavers did not show discrimination from
their responses (for sniff: T=21, N=10, P>0.250;
for physical response: T=23.5, P>0.250; and for
total response: T=27, P>0.500, Fig. 5; sign test:
C=4.5, N=10, P>0.500 for response completeness,
Our results show that beavers can discriminate
between AGS samples from unfamiliar relatives
and unfamiliar non-relatives. The long dispersal
distance, small home range and territoriality in the
beaver make it unlikely that dispersers and their
siblings have met since dispersal. We used only the
siblings that were unlikely to have met each other
before, based on large distances and land barriers
between colonies. Consequently, discrimination
between these unfamiliar siblings must be based
on self-matching (comparing the cue carried by a
test subject with that of itself) or matching with a
third individual (Holmes & Sherman 1983). This
result supports our first prediction that unfamiliar
relatives can be discriminated from unfamiliar
non-relatives. Similar discrimination by dispersers’
mates, which were familiar but did not have
any genetic relationship with the dispersers, indicates
that they can learn cues from beavers they
live with and use the cues to discriminate AGS
samples. This result provides evidence that cues
used for discrimination can be acquired indirectly
via a third individual as the reference. Therefore,
the second prediction from the phenotype
matching hypothesis is supported.
Phenotype matching has been identified as a
mechanism of kin recognition in many animal
species (e.g. Greenberg 1979; Blaustein & O’Hara
1981; Buckle & Greenberg 1981; Waldman 1981;
Grau 1982; Holmes & Sherman 1982; Getz &
Smith 1983; Holmes 1986; Brown et al. 1993).
Phenotype matching differs from social learning
by a process called cue generalization (Halpin
1991). This process allows an individual to learn a
limited number of training prototypes (often
metaphorically called labels) to form a set of
criteria (usually called a template) in the sensory
or neural system of the individual. Different individuals
are classified into different kin classes
according to the similarity between the criteria
and the cues these individuals bear (Halpin 1991).
Therefore, unfamiliar relatives carrying cues
similar to those previously encountered can be
recognized at the first encounter by this process.
Conversely, kin recognition through social learning
lacks this cue generalization process (Halpin
& Hoffman 1987); hence, a relative carrying an
unfamiliar cue will be rejected. Phenotype matching
should be the predominant mechanism in kin
recognition in situations when related individuals
are unfamiliar yet varying degrees of relatedness
need to be discriminated.
A test subject may be able to acquire the cue
about kinship used for phenotype matching in two
ways: self-learning or learning indirectly from a
third individual that is related to the test subject
(reviewed in Halpin 1991). Buckle & Greenberg
(1981) found that the kinship cue in sweat bees,
Lasioglossum zephyrum, can be acquired by selflearning
that cannot be transferred from a third
individual. Evidence about whether a cue can be
transferred from a third individual is still lacking,
however, because it is difficult to separate from
self-learning. Our second test excluded the possibility
of self-learning because the dispersers’ mates
were not related to the AGS donors. Although
this test did not examine whether beavers can
learn cues from their relatives, it does provide
indirect evidence that beavers have the ability to
do so by demonstrating that outsiders can discriminate
between strangers and individuals from
a kin group by learning the cue from one of the
kin. Since even an unrelated outsider (i.e. disperser’s
mate in our study) can learn the cue(s) used
for kin recognition within a kin group, the mechanism
of phenotype matching appears to be used
in a more general context in animal social recognition
than for kin recognition only.
Different scent sources of an individual may
convey different information (Johnston et al.
1993). Our results demonstrate a functional differentiation
between AGS and castoreum, although
both can elicit seemingly similar territorial
response (Hodgdon 1978; Welsh & Mu¨ ller-
Schwarze 1989). No evidence has been shown that
beavers use castoreum to recognize relatives
through our study. The castor sacs store castoreum
but lack secretory structures (Svendsen
1978). Many castoreum compounds are present in
the food plants of beavers (Mu¨ ller-Schwarze
1992), indicating that castoreum compounds are
derived from the environment. Although under
some specific conditions, environmentally derived
cues may be used for phenotype matching
(Gamboa et al. 1986; Porter et al. 1989), they are
considered rare (Halpin 1991). In the beaver, diet
composition changes seasonally and yearly, which
Sun & Mu¨ller-Schwarze: Sibling recognition in beaver 499
may affect the castoreum constituents. Consequently,
castoreum may not be a reliable label for
kin recognition by phenotype matching without
regular contact. In contrast, AGS is the product
of a true secretory gland (Svendsen 1978). Only
two bacterial species in low density have been
found in the gland (Svendsen & Jollick 1978), and
no evidence has shown that the bacterial contribution
to the composition of AGS compounds is
significant. Further study of AGS composition
shows that the variation over time and space is
small (Sun 1996). Therefore, it is suitable for
use in phenotype matching because the label is
It is not obvious why and how beavers reduce
their responses to the AGS samples from the
siblings of their cohabiting mates. We noticed,
however, that yearling beavers (kits) in their natal
families normally survived the replacement of one
of the adults (L. Sun, unpublished data). This
observation indicates that family reorganization
did not result in the killing of at least independent
young beavers (<2 years old) that were still living
in their parental colonies. Young beavers in their
natal families often help maintain lodges and
dams and bring food to infants (Patenaude 1984).
The loss of these beavers would mean the loss of
helpers, which may in turn reduce the reproductive
success of new adults. For this reason, it is
probably adaptive to recognize and tolerate the
offspring of new mates. Since recognition by
phenotype matching is based on the degree of
match between a label and a template, beavers
may not discriminate between the offspring
and siblings of their new mates, if the degrees of
match are the same. This may be why beavers
respond less to the unfamiliar, unrelated siblings
of their mates, although this issue needs further
Field studies of kin recognition have a few
disadvantages. It is often difficult to isolate the
variable(s) of interest from background noise.
This usually is not a problem in the laboratory.
Another problem in the field is the long time
required to wait for responses of subjects, making
it difficult to build up a reasonable sample size.
Despite these shortcomings, field studies ensure
that artificial effects such as forced choices on test
subjects are minimal, and results may be easier to
interpret in natural settings. Field studies in kin
recognition, although still scarce, have been successfuly
attempted in a few species (Waldman &
Adler 1979; Beecher et al. 1981; Waldman 1982;
O’Hara & Blaustein 1985). Our study using twosample
choice tests in the beaver has demonstrated
that, with a good understanding of the
natural history of the subjects and a careful experimental
design, hypotheses on kin recognition can
be rigorously tested in the field.
We thank Diane Chepko-Sade, Axel Engelhart,
John Kennedy, Brett Mossier, Mike Rehberg and
Boxing Wang for their assistance in our field experiment.
Meaghan Boice-Green, Diane Chepko-Sade,
Glenn Johnson and Gil Rosenthal helped us with
the manuscript preparation. We are also grateful
to the two anonymous referees for their helpful
comments and suggestions to our manuscript. This
research was supported by a Sussman Fellowship,
an Alexander Wetland Research Award, a Dence
Award Fellowship, a Friends of Moon Library
Award and two grants from the American Wildlife
Foundation to L.S. The methods have been
approved by the Institutional Committee for the
Care and Use of Animals.
Beecher, M. D., Beecher, I. M. & Lumpkin, S. 1981.
Parent–offspring recognition in bank swallows,
Riparia riparia: I & II. Anim. Behav., 29, 86–101.
Beecher, S., Hasselquist, D. & Von Schantz, T. 1994.
Genetic similarity between parents predicts hatching
failure: nonincestuous inbreeding in the great reed
warbler? Evolution, 48, 317–326.
Blaustein, A. R. 1983. Kin recognition mechanisms:
phenotype matching or recognition alleles? Am. Nat.,
Blaustein, A. R. & O’Hara, R. K. 1981. Genetic control
for sibling recognition. Nature, Lond., 290, 246–248.
Blaustein, A. R., Bekoff, M. & Daniels, J. 1987. Kin
recognition in vertebrates (excluding primates). In:
Kin Recognition in Animals (Ed. by D. J. C. Fletcher &
C. D. Michener), pp. 287–357. New York: John
Brown, G. E., Brown, J. A. & Crosbie, A. M. 1993.
Phenotype matching in juvenile rainbow trout. Anim.
Behav., 46, 123–125.
Buckle, G. R. & Greenberg, L. 1981. Nestmate recognition
in sweat bees, Lasioglossum zephyrum: does an
individual recognize its own odour or only odours of
its nestmates? Anim. Behav., 29, 802–809.
Gamboa, G. J., Reeve, H. K., Ferguson, I. D. &
Wacker, T. L. 1986. Nestmate recognition in social
wasps: the origin and acquisition of recognition
odours. Anim. Behav., 34, 685–695.
Getz, W. M. & Smith, K. B. 1983. Genetic kin recognition:
honey bees discriminate between full and half
sisters. Nature, Lond., 302, 147–148.
Grau, H. J. 1982. Kin recognition in white-footed deermice,
Peromyscus leucopus. Anim. Behav., 30, 497–
Greenberg, L. 1979. Genetic component of bee odor in
kin recognition. Science, 206, 1095–1097.
Halpin, Z. T. 1991. Kin recognition cues of vertebrates.
In: Kin Recognition (Ed. by P. G. Hepper), pp. 220–
258. Cambridge: Cambridge University Press.
Halpin, Z. T. & Hoffman, M. D. 1987. Sibling recognition
in the white-footed mouse, Peromyscus leucopus:
association or phenotype matching? Anim. Behav., 35,
Hamilton, W. D. 1964. The genetical evolution of social
behaviour, I. & II. J. theor. Biol., 7, 1–52.
Hodgdon, H. E. 1978. Social dynamics and behavior
within an unexploited beaver (Castor canadensis)
population. Ph.D. dissertation, University of
Holmes, W. G. 1986. Kin recognition by phenotype
matching in female Belding’s ground squirrels. Anim.
Behav., 34, 38–47.
Holmes, W. G. & Sherman, P. W. 1982. The ontogeny
of kin recognition in two species of ground squirrels.
Am. Zool., 22, 491–517.
Holmes, W. G. & Sherman, P. W. 1983. Kin recognition
in animals. Am. Scient., 71, 46–55.
Houlihan, P. W. 1989. Scent mounding by beaver
(Castor canadensis): functional and semiochemical
aspects. M.S. thesis, State University of New York,
College of Environmental Science and Forestry.
Jenkins, S. H. 1980. A size–distance relation in food
selection by beavers. Ecology, 61, 740–746.
Jenkins, S. H. & Busher, P. E. 1979. Castor canadensis.
Mamm. Sp., 120, 1–8.
Johnston, R. E., Derzie, A., Chiang, G., Jernigan, P. &
Lee, H. 1993. Individual scent signatures in golden
hamsters: evidence for specialization of function.
Anim. Behav., 45, 1061–1070.
Leege, T. A. 1968. Natural movements of beaver in
southeastern Idaho. J. Wildl. Mgmt, 32, 973–976.
Mu¨ ller-Schwarze, D. 1992. Castoreum of beaver (Castor
canadensis): function, chemistry and biological
activity of its components. In: Chemical Signals in
Vertebrates VI (Ed. by R. L. Doty & D. Mu¨ ller-
Schwarze), pp. 457–464. New York: Plenum Press.
Mu¨ ller-Schwarze, D. & Heckman, S. 1980. The social
role of scent marking in beaver (Castor canadensis).
J. chem. Ecol., 6, 81–95.
Mu¨ ller-Schwarze, D. & Houlihan, P. W. 1991. Pheromonal
activity of single castoreum constituents in
beaver, Castor canadensis. J. chem. Ecol., 17, 715–734.
O’Hara, R. K. & Blastein, A. R. 1985. Rana cascadae
tadpoles aggregate with siblings: an experimental field
study. Oecologia (Berl.), 67, 44–51.
Osborn, D. J. 1955. Techniques of sexing beaver, Castor
canadensis. J. Mammal., 36, 141–142.
Patenaude, F. 1984. The ontogeny of behavior of freeliving
beavers (Castor canadensis). Z. Tierpsychol., 66,
Porter, R. H., McFadyen-Ketchum, S. A. & King, G. A.
1989. Underlying bases of recognition signatures in
spiny mice, Acomys cahirinus. Anim. Behav., 37, 638–
Schulte, G. A. 1993. Chemical communication and
ecology of the North American beaver (Castor
canadensis). Ph.D. dissertation. State University of
New York, College of Environmental Science and
Schulte, G. A., Mu¨ ller-Schwarze, D. & Sun, L. 1995.
Using anal gland secretion to determine sex in beaver.
J. Wildl. Mgmt, 59, 614–618.
Sun, L. 1996. Chemical kin recognition in the beaver
(Castor canadensis): behavior, relatedness and information
coding. Ph.D. dissertation. State University of
New York, College of Environmental Science and
Svendsen, G. E. 1978. Castor and anal glands of the
beaver (Castor canadensis). J. Mammal., 59, 618–620.
Svendsen, G. E. 1980. Patterns of scent-mounding in a
population of beaver (Castor canadensis). J. chem.
Ecol., 6, 133–148.
Svendsen, G. E. 1989. Pair formation, duration of pairbonds,
and mate replacement in a population of
beavers (Castor canadensis). Can. J. Zool., 67, 336–340.
Svendsen, G. E. & Jollick, J. D. 1978. Bacterial contents
of the anal and castor glands of beaver (Castor
canadensis). J. chem. Ecol., 4, 563–569.
Townsend, J. E. 1953. Beaver ecology in western
Montana with special reference to movements. J.
Mammal., 34, 459–479.
Waldman, B. 1981. Sibling recognition in toad tadpoles:
the role of experience. Z. Tierpsychol., 56, 341–358.
Sun & Mu¨ller-Schwarze: Sibling recognition in beaver 501
Waldman, B. 1982. Sibling association among schooling
toad tadpoles: field evidence and implications. Anim.
Behav., 30, 700–713.
Waldman, B. 1987. Mechanisms of kin recognition.
J. theor. Biol., 128, 159–185.
Waldman, B. & Adler, K. 1979. Toad tadpoles associate
preferentially with siblings. Nature, Lond., 282, 611–
Walro, J. M. & Svendsen, G. E. 1981. Castor sacs and
anal glands of the North American beaver (Castor
canadensis). J. chem. Ecol., 8, 809–819.
Welsh, R. G. & Mu¨ ller-Schwarze, D. 1989. Experimental
habitat scenting inhibits colonization by beaver,
Castor canadensis. J. chem. Ecol., 15, 887–893.
502 Animal Behaviour, 54, 3