rspb.royalsocietypublishing.org
Social benefits of non-kin food sharing by
female vampire bats
Gerald G. Carter1 and Gerald S. Wilkinson2
1
Research
Cite this article: Carter GG, Wilkinson GS.
2015 Social benefits of non-kin food sharing
by female vampire bats. Proc. R. Soc. B 282:
20152524.
http://dx.doi.org/10.1098/rspb.2015.2524
Received: 20 October 2015
Accepted: 26 October 2015
Subject Areas:
behaviour
Keywords:
bet-hedging, cooperation, Desmodus rotundus,
reciprocity, social bonds, social networks
Author for correspondence:
Gerald G. Carter
e-mail: gerry@socialbat.org
Electronic supplementary material is available
at http://dx.doi.org/10.1098/rspb.2015.2524 or
via http://rspb.royalsocietypublishing.org.
2
Smithsonian Tropical Research Institute, Apartado Postal 0843-03092, Panama, Republic of Panama
Department of Biology, University of Maryland, College Park, MD, USA
Regurgitations of blood among vampire bats appear to benefit both direct
and indirect fitness. To maximize inclusive fitness, reciprocal food sharing
should occur among close kin. Why then do females with kin roost-mates
help non-kin? We tested the hypothesis that helping non-kin increases a
bat’s success at obtaining future donations by expanding its network of
potential donors. On six occasions, we individually fasted 14 adult females
and measured donations from 28 possible donors. Each female was fasted
before, during and after a treatment period, when we prevented donations
from past donors (including 10 close relatives) by simultaneously fasting
or removing them. This experiment was designed to detect partner switching and yielded three main results. First, females received less food when we
prevented donations from a past donor versus a control bat. Donors within a
group are therefore not interchangeable. Second, the treatment increased the
variance in donors’ contributions to food received by subjects, suggesting
the possibility of alternative responses to a partner’s inability to reciprocate.
Finally, bats that fed more non-kin in previous years had more donors and
received more food during the treatment. These results indicate that a bat
can expand its network of possible donors by helping non-kin.
1. Introduction
Cooperating organisms face the challenge of investing preferentially in partners
that provide the best inclusive fitness returns. Indirect fitness benefits can be
ensured through kin discrimination [1,2], while direct fitness returns can
be enforced through some combination of partner choice and partner control
(‘reciprocity’ in the broad sense [3]). However, direct and indirect fitness benefits
are often complimentary and their relative importance can shift over evolutionary
time. Cooperation might first arise through kin selection, but later become stabilized by direct fitness benefits [4,5]. We hypothesize that this scenario explains
regurgitated food sharing in the common vampire bat (Desmodus rotundus) [6].
Vampire bats are obligate blood-feeders on a tight energy budget. Stable social
networks of eight to 12 adult females arise despite frequent switching among two
to five roosts [6]. Frequent roost-mates regurgitate blood to kin and non-kin in dire
need due to failed foraging (mean kinship per group: 0.03–0.11) [6]. Past work on
predictors of food sharing [6–8] suggest that these donations may have originated
as maternal care and were co-opted for helping adult kin, and then co-opted again
for promoting reciprocal donations. If vampire bat food donations do indeed provide both direct and indirect benefits, then the largest possible inclusive fitness
benefits should accrue from reciprocal sharing with close kin. Why then do
females with kin roost-mates help non-kin?
Non-kin donations may allow bats to create and maintain a wider network
of potential donors than would be available if food sharing was limited to close
kin. This could occur with or without direct reciprocity. In the absence of
reciprocity and assuming that bats exhibit no dyadic preferences when aiding
non-kin groupmates, non-kin sharing might simply help sustain a larger
group of possible donors by increasing survival of roost-mates (via passive
group augmentation [9– 11]). Alternatively, cooperation might be enforced by
partner choice, partner control or both [3,12–16]. This scenario assumes that
bats invest in specific social bonds based on the fitness-relevant returns, and
it predicts that bats which make better social investments will be more
& 2015 The Author(s) Published by the Royal Society. All rights reserved.
baseline
2
14
1
S
13
9
12
4
6
10 11
2
14
8
5
1
S
2
9
7
12
3
4
6
X
13
6
10 11
2
14
1
S
8
13
X
9
12
4
8
10 11
5
1
7
3
2
post-treatment
14
S
13
X
9
7
3
5
*
treatment
7
12
3
4
6
8
10 11
successful even within a group of constant size. In other
words, non-kin sharing widens social network size not
merely by increasing groupmate survival, but by creating
or strengthening social ties that yield reciprocal returns. As
a consequence, individuals that feed more non-kin should
have more donors when the need arises.
We tested this prediction in a captive group of vampire
bats. We experimentally prevented food sharing by simultaneous fasting or temporarily removing a partner with a
prior history of food sharing. We tested three hypotheses. If
donors within the group are not interchangeable, then preventing donations from past donors, rather than control bats,
would lead to an immediate decline in total food received. If
non-kin investments provide long-term cooperative returns,
then bats that fed more non-kin in previous years would
have more donors and be fed more when targeted kin
donors could not feed them. Finally, if an inability to reciprocate leads to rapid partner switching, then after a subject
repeatedly fails to feed a past donor, a lower proportion of a
subject’s received food would come from that donor.
2. Material and methods
(a) General methods
Animal care and methods for inducing food regurgitations
are described elsewhere [7]. We housed 24 – 39 captive-born
D. rotundus, sourced from three different zoos, which led to multiple matrilines and unrelated bats, as found in wild colonies [6].
Durations of regurgitations in these captive bats are similar to
those observed in the wild [17]. To induce and quantify food
sharing, we removed and fasted two subjects for 24 h, returned
the first subject to the group for 1 h of video focal sampling,
and then repeated this with the second subject. To measure
food sharing, we scored seconds of time that the subject licked
the mouth of a donor. Mouth-licking is a good measure of
food sharing because it is highly correlated with trial weight
gain, but it is not affected by urination and defaecation (see the
electronic supplementary material, figure S1).
We used data on food sharing spanning the years 2010–2014
(approx. 1250 regurgitation observations, 355 trial-donations, 442
directed links between dyads, from [7] and afterwards). We
measured two basic forms of social network centrality: the
number of bats that have fed a subject (‘indegree’) and the
number of bats that a subject has fed (‘outdegree’). We estimated
relatedness using the R package ‘related’ [18] from genotypes of
19 polymorphic microsatellite markers (see the electronic supplementary material for details). We define ‘non-kin’ as partners
with a pairwise relatedness estimate of zero or less.
(b) Partner-switching experiment
In this experiment, we quantified food sharing before, during and
after a treatment period. During this treatment, we inhibited food
sharing between members of seven targeted dyads, then tested for
subsequent partner switching, i.e. that a targeted past donor contributed significantly less than before to feeding the fasted
subject (figure 1). The targeted dyads were 14 female subjects
paired according to their history of food sharing (electronic supplementary material, table S3). We paired as many females as
possible with their most frequent donor, which resulted in two
pairs of non-kin, four mother–adult daughter pairs and one
other kin pair (electronic supplementary material, table S3). We
focused on testing females because they perform about 80% of
food donations [7], and we targeted frequent donors to maximize
our ability to detect a decrease in donation rate.
We tested each subject six times for a total of 84 fasting trials.
We conducted two trials each trial day; one round of testing all
14 subjects lasted one week. For each trial, we fasted the subject,
then measured any donations from bats in the group. A subject
could potentially be fed by 13 females and 11 – 15 males, but
one female (the other subject that day) was fasted on the same
night and was therefore either missing or unfed. We grouped
these ‘missing’ and ‘unfed’ conditions together because we did
not detect any difference in how bats responded to these two
ways of preventing donations (see the electronic supplementary
material). During the first two-week pre-treatment period, we prevented donations from a control bat that had not previously fed
the subject much or at all. During the next three-week treatment
period, we prevented donations from the targeted donor. In
the final post-treatment week, we again prevented donations
from a control bat. We predicted that the treatment would:
(i) reduce the amount of food received by subjects, (ii) reveal
that bats that fed more non-kin in previous years would have
more donors and receive more food when their primary
donors could not feed them, and (iii) reduce food sharing with
targeted donors relative to other donors.
(c) Statistical analyses
Before conducting any parametric test, we tested for deviations
from normality using a goodness of fit test and assumed unequal
variances for all t-tests. Whenever parametric assumptions were
violated, we used permutation tests (lmPerm package in R). To
help interpret null results, we provide 95% confidence intervals.
We accounted for repeated measures by testing before versus
after contrasts by bat.
To examine the effect of prior non-kin donations on sharing
success in the experiment, we first used permuted simple linear
regressions to test whether the number of male or female nonkin recipients in prior years predicted: (i) number of donors in
prior years, (ii) number of donors during the experiment, or
Proc. R. Soc. B 282: 20152524
Figure 1. Illustration of how the partner-switching experiment was conducted for a representative individual. Circles are female bats. Arrows show food given.
Fasted bats are labelled blue. Each of 14 subjects (S) started the experiment with a baseline history of sharing with several prior donors (red) ranked by donation
rate (number). For each subject, we targeted a unique donor of highest possible rank (*). In treatment trials, we prevented donations from this targeted donor. In
pre- and post-treatment trials, we prevented donations from a control bat. We measured food donations in each trial. (Online version in colour.)
rspb.royalsocietypublishing.org
5
pre-treatment
total food received (s) during treatment
1500
1000
500
0
0
2
4
6
8
no. non-kin females fed
10
Figure 2. Donations to unrelated females predict success when primary
donors are removed. The number of non-kin females previously fed by a subject during 2010– 2014 predicts the total food received during the
experimental treatment period. Model fit is improved by controlling for
the number of times the subject was tested (circle size, see the electronic
supplementary material). (Online version in colour.)
800
700
600
food received (s)
Females that previously fed more non-kin had more donors
and they received more food. The number of non-kin fed
by a female in previous years predicted the average number
of donors per experimental trial (rank-transformed response:
R 2 ¼ 0.43, p ¼ 0.008). More donors per trial led to more
food received overall (Spearman’s rank correlation: r ¼ 0.59,
p ¼ 0.025). The total food received by a female during treatment trials was predicted by its prior propensity to feed nonkin females (R 2 ¼ 0.48, p ¼ 0.006), but not non-kin males
(R 2 ¼ 0.12, p ¼ 0.23), related males (R 2 ¼ 0.02, p ¼ 0.60) or
related females (R 2 ¼ 0.07, p ¼ 0.38). Stepwise regression
chose ‘number of female non-kin previously fed’ as the
chief determinant of sharing success while excluding other
numbers of previously fed bats. Minimum BIC also selected
‘times tested’, so we included it in the model (adj. R 2 ¼ 0.60,
F1,1 ¼ 11.8, p ¼ 0.0064; figure 2). The model confirmed that
sharing success was predicted by the number of previously
fed female non-kin ( p ¼ 0.002) but not the number of times
the bat was tested in prior years ( p ¼ 0.09).
Targeted donors were not immediately replaced by other
groupmates in a trial. Subjects received less food when their
targeted donor, rather than a control bat, was removed
from the donor pool ( paired t13 ¼ 22.7, one-sided p ¼
0.0089). A decline in food received (mean ¼ 272 s, 95% confidence interval (CI) ¼ 55 –489 s) was detectable only when the
targeted donor was first removed and not during other fasting rounds. During this same time, we did not detect a
change in the number of donors (mean ¼ 20.36, 95%
3
500
400
300
200
100
0
pre
during
post
Figure 3. Mean food received during experiment. Mean and 95% CIs for
seconds of food received in total (black) and from targeted donors (grey)
during the pre-treatment, treatment and post-treatment periods.
CI ¼ 20.84 to 0.13 bats). Across all trials, bats were fed by
an average of 2.6 donors (95% CI ¼ 2.3–2.9).
After three weeks with no sharing between targeted partners, the mean food given by targeted donors did not
decrease (figure 3; mean ¼ 220.68 s, 95% CI ¼ 2177 to
þ136 s, t10 ¼ 20.29, p ¼ 0.77) nor did the donor’s contribution
to the subject’s total food received (mean ¼ þ18%, 95%
CI ¼ 234% to þ70%, paired t-test with arcsinetransformed proportions: t10 ¼ 0.78, p ¼ 0.45); however, this
lack of a difference was not due to sharing rebounding to
near its original level across all donors. Rather, after the treatment donors tended to provide either a much higher or
lower percentage of their partner’s total food gain, i.e. a
bimodal response, with a higher variance in donor contributions to total food received (figure 4; Levene’s test:
F1,31 ¼ 17.4, p ¼ 0.0002). The donor’s contributions to food
Proc. R. Soc. B 282: 20152524
3. Results
2000
rspb.royalsocietypublishing.org
(iii) total food received across all treatment trials. To select the best
multivariate model for predicting total food received during treatment, we entered six predictors into backwards and forwards
stepwise regression and evaluated model fit using minimum
Akaike and Bayesian information criterion (AIC and BIC). These
predictors were age, the number of times the bat was fasted
(which is proportional to the number of times they could have
donated), and the numbers of previously fed male kin, male
non-kin, female kin and female non-kin. We used Spearman’s correlation to confirm that females with more donors per trial received
more food.
To determine if targeted donor contributions were replaceable,
we used a paired t-test to determine if a subject’s total food received
declined during the first treatment trial, specifically when
donations from the targeted donor were first prevented. We also
tested if a decline in food received was detectable during any
other trial rounds. We repeated this analysis for the number of
donors per trial. Next, to determine if the bats showed a response
to a past donor’s inability to reciprocate, we first compared the targeted donor’s contribution to food received in the pre-treatment
and post-treatment periods. We tested both absolute values of
donation sizes and arcsine-transformed proportions of subjects’
total food received. Three subjects (from two dyads) were not fed
by their targeted donors during pre-treatment and were therefore
excluded from this analysis (electronic supplementary material,
table S3); however, the main result was consistent whether we
included these cases or not.
Finally, we tested if the targeted donor’s contribution to a
subject’s total food received became more variable after the treatment period. We conducted Levene’s test and O-Brien’s test on
the arcsine-transformed percentages, as well as a permutation
test (see the electronic supplementary material). Results were
consistent across all tests, so we present only the results of the
first test we conducted.
75%
50%
0%
pre
post
Figure 4. Effect of treatment on proportions of food received from targeted
donor. Changes in the targeted donor’s contribution to total food received
from the pre-treatment period to the post-treatment period. Red lines
show bats paired with maternal kin targeted donors. Blue lines show bats
paired with non-kin targeted donors. Electronic supplementary material,
figure S3, shows both rounds of the pre-treatment. (Online version in colour.)
received decreased to 0% in five of 11 cases and increased to
near 100% in four cases (figure 4; electronic supplementary
material, figure S3).
4. Discussion
Females that had previously fed more non-kin females were
subsequently more successful at obtaining food when we prevented sharing from a primary donor (such as a mother or
adult daughter). Prior donations to female non-kin, a specific
form of network centrality, predicted the ability of a female
bat to obtain later donations. Food-sharing bonds are individualized, stable [7,8] and non-interchangeable, such that food
from a primary donor cannot be quickly replaced by food
from other group members when that donor is absent or
unable to help. Consequently, the quantity of these foodsharing bonds can influence sharing success on both short
and long timescales even within a group of stable size.
The social rewards of food sharing are evident even
before taking into account any potential for passive group
augmentation through increased group mate survivorship.
However, this experiment failed to demonstrate a consistent
contingent response within highly bonded dyads. Not all
targeted donors reduced their investments to zero after
the treatment period. Instead, some donors increased their
contributions leading to a bimodal distribution of outcomes
(figure 4). This pattern is consistent with the possibility of
divergent strategies for responding to an inability to reciprocate. A significant decrease in a donor’s contribution could
indicate partner switching, whereas a significant increase in
a donor’s contribution could indicate a form of relationship
repair. More experimental trials with kin and non-kin are
needed to test this possibility.
Food sharing in vampire bats is commonly cited as an
example of reciprocity because non-kin dyads appeared to
exchange blood in a simple tit-for-tat like pattern [6]. However,
4
Proc. R. Soc. B 282: 20152524
25%
the results from this experiment show that some food-sharing
bonds are not based on strict contingent turn-taking and persist
beyond three weeks without reciprocal exchanges. Conclusions
regarding contingent reciprocity should be drawn carefully
with several factors in mind. First, close kinship may reduce
contingency. Second, if contingency is subtle or gradual,
then more dyads and more measures over time are needed
to achieve the necessary statistical power. We have observed
that previously stable food-sharing relationships changed
when three bats were moved into a larger colony with more
potential partners (G. G. Carter 2014, unpublished data).
Hence, partner switching occurs with dramatic changes in
group composition. However, partner switching based
on food-sharing experiences probably require more time,
given that new food-sharing bonds appear to develop
gradually [8]. Third, subjects may respond less severely to
non-reciprocating partners that are absent or unfed, such as
in this experiment, compared to cases where partners have
food but still reject solicitations (see further discussion of this
issue in the electronic supplementary material). This ‘excuse
principle’ has been demonstrated in cooperatively mobbing
pied flycatchers [19] and cooperatively breeding cichlid fish
[20]. Finally, allogrooming in vampire bats is more frequent
than in other bats [21] and linked to food sharing both by
occurrence [7] and a common neuroendocrine mechanism
[22]. Allogrooming may therefore compensate for imbalances
in food sharing, as shown in primates [23].
Vampire bat food sharing deviates from many assumptions of simple cooperation models based on the iterated
Prisoner’s Dilemma [4,5]. These models have had a major
impact on subsequent social evolution theory, but literal translations overemphasize short-term contingent exchanges of a
single service with a single partner, and this view constrains
how researchers have tested cooperation [3]. Increasing evidence suggests that mechanisms for maintaining cooperation
depend on factors missing from the iterated Prisoner’s
Dilemma model, such as partner choice, continuous (rather
than binary) investments and multiple cooperative services
(e.g. grooming, sharing or alliances [3]). A stable social bond
with diverse benefits might require only a single cognitive
index of relationship quality rather than separate accounting
of all prior cooperative investments and returns, allowing for
more subtle contingencies [3,23–25]. For example, short-term
contingency is weakest in strongly bonded primates and more
evident in weakly bonded partners [24]. In vampire bats, each
dyadic relationship might involve multiple cooperative returns,
including indirect fitness benefits [6], allogrooming [7] and
social thermoregulation [26]. Testing alternative mechanisms
for maintaining cooperation will therefore require carefully
manipulating the cooperative value of specific partners while
measuring cooperative investments across different partners
(e.g. [21]).
Reciprocity is most clear in non-kin without social bonds
or partner choice [3], but in nature, reciprocity might often
occur among kin with multiple social bonds. In our experiments, the total cost of feeding each hungry bat was often
divided among several donors. Even bats with a mother or
adult daughter present still received about half of their
food from other donors (electronic supplementary material,
figure S2). By increasing their sharing networks, bats can
also reduce the costs incurred by each of their donors. This
has interesting potential consequences for the cost-benefit
ratios of helping. For example, if bats with more sharing
rspb.royalsocietypublishing.org
% food sharing from primary donor
100%
Ethics. All procedures were approved by the University of Maryland
Institutional Animal Care and Use Committee ( protocol R-10–63).
Data accessibility. The data supporting this article have been posted to
Figshare.com or uploaded as part of the electronic supplementary
material. DNA sequences: GenBank accession no. PRJNA279293.
Authors’ contributions. G.G.C. conceived of and designed the study, carried out the molecular laboratory work, data collection and statistical
analysis, and drafted the manuscript; G.S.W. advised the study
design, data analysis and writing. Both authors gave final approval
for publication.
Competing interests. We have no competing interests.
Funding. This work was supported by a Ford Predoctoral Fellowship, a
Dissertation Improvement Grant from the National Science Foundation and grants from the American Society of Mammalogists and
Animal Behavior Society.
Acknowledgements. We thank the Organization for Bat Conservation for
their extraordinary support. K. Carleton, G. Borgia, K. Thompson,
M. Taborksy and three anonymous reviewers provided feedback
that improved the manuscript.
1.
Griffin AS, West SA. 2003 Kin discrimination and the
benefit of helping in cooperatively breeding
vertebrates. Science 302, 634–636. (doi:10.1126/
science.1089402)
2. Hamilton WD. 1964 The genetical evolution of
social behavior. J. Theor. Biol. 7, 1–51. (doi:10.
1016/0022-5193(64)90038-4)
3. Carter GG. 2014 The reciprocity controversy.
Anim. Behav. Cogn. 1, 368 –386. (doi:0.12966/abc.
08.11.2014)
4. Trivers RL. 1971 The evolution of reciprocal altruism.
Q. Rev. Biol. 46, 35 –57. (doi:10.1086/406755)
5. Axelrod R, Hamilton WD. 1981 The evolution of
cooperation. Science 211, 1390–1396. (doi:10.
1126/science.7466396)
6. Wilkinson GS. 1984 Reciprocal food sharing in the
vampire bat. Nature 308, 181 –184. (doi:10.1038/
308181a0)
7. Carter GG, Wilkinson GS. 2013 Food sharing in
vampire bats: reciprocal help predicts donations
more than relatedness or harassment. Proc. R.
Soc. B 280, 20122573. (doi:10.1098/rspb.
2012.2573)
8. Carter G, Wilkinson G. 2013 Does food sharing in
vampire bats demonstrate reciprocity? Commun.
Integr. Biol. 6, e25783. (doi:10.4161/cib.25783)
9. Kokko H, Johnstone RA, Clutton-Brock TH. 2001 The
evolution of cooperative breeding through group
augmentation. Proc. R. Soc. Lond. B 268, 187–196.
(doi:10.1098/rspb.2000.1349)
10. Kingma SA, Santema P, Taborsky M, Komdeur J.
2014 Group augmentation and the evolution of
cooperation. Trends Ecol. Evol. 29, 476 –484.
(doi:10.1016/j.tree.2014.05.013)
11. Foster KR. 2004 Diminishing returns in social
evolution: the not-so-tragic commons. J. Evol. Biol.
17, 1058–1072. (doi:10.1111/j.1420-9101.2004.
00747.x)
12. Noë R, Hammerstein P. 1994 Biological markets:
supply and demand determine the effect of partner
choice in cooperation, mutualism and mating.
Behav. Ecol. Sociobiol. 35, 1 –11. (doi:10.1007/
BF00167053)
13. Rutte C, Taborsky M. 2008 The influence of social
experience on cooperative behaviour of rats (Rattus
norvegicus): direct vs generalised reciprocity. Behav.
Ecol. Sociobiol. 62, 499 –505. (doi:10.1007/s00265007-0474-3)
14. Krams I, Krama T, Igaune K, Mand R. 2008
Experimental evidence of reciprocal altruism
in the pied flycatcher. Behav. Ecol. Sociobiol. 62,
599 –605. (doi:10.1007/s00265-007-0484-1)
15. Kiers ET et al. 2011 Reciprocal rewards stabilize
cooperation in the mycorrhizal symbiosis. Science
333, 880 –882. (doi:10.1126/science.1208473)
16. Dolivo V, Taborsky M. 2015 Norway rats reciprocate
help according to the quality of help they
received. Biol. Lett. 11, 20140959. (doi:10.1098/
rsbl.2014.0959)
17. Wilkinson GS, Carter GG, Bohn KM, Adams DM.
2015 Non-kin cooperation in bats. Phil. Trans. R.
Soc. B 371, 20150095. (doi:10.1098/rstb.2015.0095)
18. Pew J, Muir PH, Wang J, Frasier TR. 2015 related:
an R package for analysing pairwise relatedness
from codominant molecular markers. Mol. Ecol. Res.
15, 557– 561. (doi:10.1111/1755-0998.12323)
19. Krams I, Kokko H, Vrublevska J, Abolins-Abols M,
Krama T, Rantala MJ. 2013 The excuse principle
20.
21.
22.
23.
24.
25.
26.
can maintain cooperation through forgivable
defection in the Prisoner’s Dilemma game.
Proc. R. Soc. B 280, 20131475. (doi:10.1098/
rspb.2013.1475)
Fischer S, Zöttl M, Groenewoud F, Taborsky B. 2014
Group-size-dependent punishment of idle
subordinates in a cooperative breeder where helpers
pay to stay. Proc. R. Soc. B 281, 20140184. (doi:10.
1098/rspb.2014.0184)
Carter GG, Leffer LL. 2015 Social grooming in bats:
are vampire bats exceptional? PLoS ONE 10,
e0138430. (doi:10.1371/journal.pone.0138430)
Carter GG, Wilkinson GS. In press. Intranasal
oxytocin increases social grooming and food sharing
in the common vampire bat Desmodus rotundus.
Horm. Behav. (doi:10.1016/j.yhbeh.2015.10.006)
Fruteau C, Voelkl B, van Damme E, Noë R. 2009
Supply and demand determine the market value of
food providers in wild vervet monkeys. Proc. Natl
Acad. Sci. USA 106, 12 007– 12 012. (doi:10.1073/
pnas.0812280106)
Seyfarth RM, Cheney DL. 2012 The evolutionary
origins of friendship. Annu. Rev. Psychol. 63,
153–177. (doi:10.1146/annurev-psych-120710100337)
Schino G, Aureli F. 2009 Reciprocal altruism in
primates: partner choice, cognition, and emotions.
Adv. Stud. Behav. 39, 45 –69. (doi:10.1016/S00653454(09)39002-6)
Delpietro VH, Russo RG. 2002 Observations of the
common vampire bat (Desmodus rotundus) and the
hairy-legged vampire bat (Diphylla ecaudata) in
captivity. Z. Säugetierkd. 67, 65 –78. (doi:10.1078/
1616-5047-00011)
Proc. R. Soc. B 282: 20152524
References
5
rspb.royalsocietypublishing.org
partners require less help from each donor, they might be
preferred social partners. This might help explain why bats
sometimes strongly prefer particular non-kin partners to the
point of rejecting initial donation offers from others [7].