Lethal Aggression in Mobile Forager Bands and Implications for the Origins of War

Abstract

It has been argued that warfare evolved as a component of early human behavior within foraging band societies. We investigated
lethal aggression in a sample of 21 mobile forager band societies (MFBS) derived systematically from the standard cross-cultural
sample. We hypothesized, on the basis of mobile forager ethnography, that most lethal events would stem from personal disputes
rather than coalitionary aggression against other groups (war). More than half of the lethal aggression events were perpetrated
by lone individuals, and almost two-thirds resulted from accidents, interfamilial disputes, within-group executions, or interpersonal
motives such as competition over a particular woman. Overall, the findings suggest that most incidents of lethal aggression
among MFBS may be classified as homicides, a few others as feuds, and a minority as war.

Full text

DOI: 10.1126/science.1235675
, 270 (2013);341 Science
Douglas P. Fry and Patrik Söderberg
Origins of War
Lethal Aggression in Mobile Forager Bands and Implications for the
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by
, you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines
can be obtained byPermission to republish or repurpose articles or portions of articles
): December 11, 2013 www.sciencemag.org (this information is current as of
The following resources related to this article are available online at
http://www.sciencemag.org/content/341/6143/270.full.html
version of this article at:
including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/content/suppl/2013/07/17/341.6143.270.DC1.html
can be found at: Supporting Online Material
http://www.sciencemag.org/content/341/6143/270.full.html#related
found at:
can berelated to this article A list of selected additional articles on the Science Web sites
http://www.sciencemag.org/content/341/6143/270.full.html#ref-list-1
, 1 of which can be accessed free:cites 17 articlesThis article
http://www.sciencemag.org/cgi/collection/anthro
Anthropology
subject collections:This article appears in the following
registered trademark of AAAS.
is aScience2013 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on December 11, 2013www.sciencemag.orgDownloaded from on December 11, 2013www.sciencemag.orgDownloaded from on December 11, 2013www.sciencemag.orgDownloaded from on December 11, 2013www.sciencemag.orgDownloaded from on December 11, 2013www.sciencemag.orgDownloaded from

and Weddell Seas (29, 31–33) (see supplemen-
tary materials), but total melting of 10 of the
larger ice shelves is notably less here than in
circumpolar models (7, 11). Discrepancies be-
tween model results and observations have
been attributed to deficiencies in atmospheric
forcing, the representation of sea-ice cover, the
smoothing of bottom topography, and assump-
tions regarding cavity shape. Some models yield
annual cycles and decadal variability (29)that
can now be compared for specific periods with
glaciological measurements, which need to be
extended in time.
Our results indicate that basal melting accounts
for a larger fraction of Antarctic ice-shelf attrition
than previously estimated. These improved glaci-
ological estimates provide not only more accurate
and detailed reference values for modeling but
also a baseline for similar future studies. Ice-shelf
meltwater production exhibits a complex spatial
pattern around the continent, with an outsized
contribution of smaller , fast-melting ice shelves
in both West and East Antarctica. W a rm-cavity ice
shelves along the Southeast Pacific coastline, pre-
dicted and observed to be sensitive to ocean warm-
ing and circulation strength (9, 34), were thinning
rapidlyin2003to2008(23). Nearly half of the East
Antarctic ice shelves were also thinning, some due
to probable exposure to “warm” seawater, with
connections to ice drainage basins grounded below
sea level.
Continued observations of ice-shelf velocity
and thickness change, along with more detailed
information on cavity shape, seafloor topography,
and atmospheric and oceanic forcing variability
are critical to understand the temporal variability
and evolution of Antarctic ice shelves. Continued
warming of the ocean will slowly increase ice-
shelf thinning, but if major shifts in sea ice cover
and ocean circulation tip even large ice-shelf
cavities from cold to warm (35), there could be
major changes in ice shelf and thus ice-sheet
mass balance.
References and Notes
1. P. Fretwell et al., The Cryosphere 7, 375–393 (2013).
2. C. W. Swithinbank, Satellite Image Atlas of Glaciers of the
World: Antarctica, R. S. Williams, J. G. Ferrigno, Eds.
(USGS Prof. Paper 1386-B, 1988).
3. N. I. Barkov, Ice Shelves of Antarctica (New Delhi, NY,
Amerind Publishing Company, 1985).
4. R. LeB. Hooke, Principles of Glacier Mechanics
(Cambridge University Press, Cambridge, 2005).
5. K. M. Cuffey, W. S. B. Paterson, The Physics of Glaciers
(Elsevier, Burlington, MA, ed. 4, 2010).
6. S. S. Jacobs et al., J. Glaciol. 38, 375 (1992).
7. H. H. Hellmer, Geophys. Res. Lett. 31, L10307 (2004).
8. A. Jenkins, S. S. Jacobs, J. Geophys. Res. 113, (C4),
C04013 (2008).
9. S. S. Jacobs, A. Jenkins, C. F. Giulivi, P. Dutrieux, Nature
Geosci. 4, 519–523 (2011).
10. A. Foldvik, T. Gammelsrod, E. Nygaard, S. Osterhus,
J. Geophys. Res. Oceans 106, 4463 (2001).
11. R. Timmermann, Q. Wang, H. H. Hellmer, Ann. Glaciol.
53, 303–314 (2012).
12. I. Joughin, L. Padman, Geophys. Res. Lett. 30, 1477
(2003).
13. J. Wen et al., J. Glaciol. 56,81–90 (2010).
14. E. Rignot, S. S. Jacobs, Science 296, 2020–2023
(2002).
15. A. Jenkins, C. S. M. Doake, J. Geophys. Res. 96, 791
(1991).
16. J. T. M. Lenaerts et al., J. Geophys. Res. 117, D05108
(2012).
17. E. Rignot et al., Nat. Geosci. 1, 106–110 (2008).
18. C. Allen, IceBridge MCoRDS L2 Ice Thickness. Boulder,
Colorado, USA: NASA DAAC at the National Snow and Ice
Data Center (2010).
19. D. D. Blankenship, S. Kempf, D. Young, IceBridge
HiCARS 2 L2 Geolocated Ice Thickness. Boulder, Colorado,
USA: NASA DAAC at the National Snow and Ice Data
Center (2012).
20. J. A. Griggs, J. L. Bamber , J. Glaciol. 57, 485–498
(2011).
21. E. Rignot, J. Mouginot, B. Scheuchl, Geophys. Res. Lett.
38, L10504 (2011).
22. E. Rignot, J. Mouginot, B. Scheuchl, Science 333,
1427–1430 (2011).
23. H. D. Pritchard et al., Nature 484, 502–505 (2012).
24. K. Makinson et al., Geophys. Res. Lett. 38, L06601
(2011).
25. H. J. Horgan, R. T. Walker, S. Anandakrishnan,
R. B. Alley, J. Geophys. Res. 116, C02005 (2011).
26. A. Shepherd et al., Science 338 , 1183–1189 (2012).
27. S. Neshyba, E. G. Josberger, J. Phys. Oceanogr. 10,
1681–1685 (1980).
28. K. Grosf eld et al., Antarct. Res. Ser. 75, 319–339 (1998).
29. M. P. Schodlok, D. Menemenlis, E. Rignot, M. Studinger ,
Ann. Glaciol. 53, 156–162 (2012).
30. G. D. Williams et al., Deep Sea Res. Part II Top. Stud.
Oceanogr. 58, 1194–1210 (2011).
31. P. R. Holland, A. Jenkins, D. Holland, Geophys. Res. Lett.
115, (C5), C05020 (2010).
32. L. Padman et al., J. Geophys. Res. 117, C01010
(2012).
33. P. R. Holland et al., Geophys. Res. Lett. 36, L11604
(2009).
34. P. R. Holland, A. Jenkins, D. M. Holland, J. Clim. 21,
2558–2572 (2008).
35. H. H. Hellmer et al., Nature 485, 225–228 (2012).
Acknowledgments: We thank three anonymous reviewers for
their constructive criticism of the manuscript. This work was
performed at the University of California, Irvine, and at the Jet
Propulsion Laboratory, California Institute of Technology, under
grants from NASA’s Cryospheric Science Program and
Operation IceBridge (OIB) and at the Lamont-Doherty Earth
Observatory of Columbia University under grants from the
National Science Foundation and the National Oceanic and
Atmospheric Administration.
Supplementary Materials
www.sciencemag.org/cgi/content/full/science.1235798/DC1
Supplementary Text
Figs. S1 to S4
Tables S1 and S2
References (36–58)
29 January 2013; accepted 31 May 2013
Published online 13 June 2013;
10.1126/science.1235798
Lethal Aggression in Mobile
Forager Bands and Implications
for the Origins of War
Douglas P. Fry
1,2
* and Patrik Söderberg
1,3
It has been argued that warfare evolved as a component of early human behavior within
foraging band societies. We investigated lethal aggression in a sample of 21 mobile forager band
societies (MFBS) derived systematically from the standard cross-cultural sample. We hypothesized,
on the basis of mobile forager ethnography, that most lethal events would stem from personal
disputes rather than coalitionary aggression against other groups (war). More than half of the
lethal aggression events were perpetrated by lone individuals, and almost two-thirds resulted
from accidents, interfamilial disputes, within-group executions, or interpersonal motives such as
competition over a particular woman. Overall, the findings suggest that most incidents of
lethal aggression among MFBS may be classified as homicides, a few others as feuds, and a
minority as war.
A
controversy exists regarding mobile for-
ager band societies (MFBS) and war-
fare. Field researchers who have worked
with MFBS generally report that warfare is ab-
sent or rudimentarily developed (1–6). Fry (7)
compared MFBSs with complex and equestri-
an foragers and found that most MFBS (62%)
were nonwarring, whereas all of the complex
and equestrian societies made war. On the other
hand, Wrangham and Glowacki [(8), p. 7] de-
veloped a chimpanzee-based lethal raiding mod-
el, asserting that “humans evolved a tendency
to kill members of other groups,” andtheypro-
vided ethnographic quotations on MFBS to il-
lustrate the model. They [(8), p. 8] define war as
when “coalitions of members of a group seek to
inflict bodily harm on one or more members of
another group; ‘groups’ are independent politi-
cal units.” Bowles (9) examined war deaths in
eight societies , six of which were MFBS, and re-
ported the occurrence of war in all eight societies,
which he takes as confirmation that war has been
pervasive during human evolution.
1
Peace, Mediation and Conflict Research, Åbo Akademi Uni-
versity in Vasa, Post Office Box 311, FIN-65101, Vasa, Finland.
2
Bureau of Applied Research in Anthropology, School of An-
thropology, Post Office Box 210030, Tucson, AZ 85721–0030,
USA.
3
Developmental Psychology, Åbo Akademi University in
Vasa, Post Office Box 311, FIN-65101, Vasa, Finland.
*Corresponding author. E-mail: dfry@abo.fi
19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org270
REPORTS

There are two likely explanations for such di-
vergent interpretations about warfare and MFBS:
differences in how warfare is defined and dif-
ferences in sampling procedure and composition
(7, 8). Recognizing that definitional opinions
differ, rather than making an a priori determi-
nation regarding which events are classified as
manslaughter , homicide, feud, or war, we instead
consider the salient characteristics of each and
every actual event involving lethal aggression in
a systematically derived, representative sample
of MFBS.
There are numerous reasons to predict a
paucity of warfare among MFBS (see supple-
mentary materials). (i) In MFBS, group size is
too small to support warfare. (ii) MFBS have
egocentric social networks with descent gen-
erally figured bilaterally through both parental
lines. This does not facilitate coalitional alliance
formation useful in war. (iii) Group composition
fluctuates over time, resulting in kinship and so-
cial networks that cut across different groups,
a factor that dampens intergroup hostility. (iv)
MFBS tend not to be segmented into subgroups.
In terms of residence, MFBS tend to be multi-
local, not patrilocal, thus lacking a residential
pattern known to facilitate coalitions among
male kin useful for war. (v) The social order is
egalitarian and leadership is lacking; no one
has the authority to order others to fight. (vi)
Foraging areas are large, population densities
low , and resources spread-out, making defense
of territory difficult or impossible. (vii) Bound-
aries often are controlled socially through use-
requests and permission-granting. (viii) Typical
spoils of war—material goods or stored food—
are largely lacking, and the necessity of mobility
makes the capture and containment of individ-
uals against their will (e.g., slaves or brides) im-
pra ctical (and runs counter to the MFBS ethos
of egalitarianism). (ix) Conflicts within and be-
tween groups are easily handled by separation
and other conflict-resolution mechanisms. On
the basis of these foregoing characteristics, we
would expect lethal aggression in MFBS to be
mostly interpersonal, not intergroup. Additional-
ly, in mammals the killing of conspecifics is an
atypical and infrequent form of aggression com-
pared to displays, noncontact threats, and restrained
aggr ession, so perhaps also for humans the de-
velopment of an evolutionary model based on
restraint as a widely documented phenomenon
across species, rather than on rare killing behav-
ior , merits consideration (10).
We extracted a subsample of purely MFBS
(n = 21) from the standard cross-cultural sample
(SCCS). T o circumvent sampling bias , ra th er tha n
self-selecting cases, we derived the sample of
MFBS based on the published rating crit e r i a of
others researchers (11, 12). During data collection,
we used only the principal authority sources
(PAS) as the earliest, high-quality ethnographic
descriptions available (12). We considered every
instance of lethal aggression reported for these
21 MFBS (13).
The 21 MFBS produced a total of 148 lethal
aggression events. The median number was 4
(mean = 7.05; SD = 14.64), with a range from
0 to 69. One society , the Tiwi of Australia, had
an exceptionally large number of lethal events
(n = 69). If the Tiwi case is removed, the me-
dian number of lethal events for the remaining
20 societies drops to 3.5, the mean is almost
cut in half (mean = 3.95; SD = 3.69), and the
range is reduced to 0 to 15.
Of 135 lethal events with unambiguous per-
petrator and victim information, 55% consisted
of one killer and one victim. In 23% of the le-
thal events, more than one person participated
in killing a single individual, and in 22% of
the events, more than one person participated
in killing more than one person (Fig. 1). In only
one lethal event (0.7%), did a single killer dis-
patch more than one person (table S4, case
18), and the two victims were children. Tiwi so-
ciety reflects a different pattern wherein 44%
of the lethal events involved one killer and one
victim, whereas the corresponding figure for the
other 20 societies combined was 64% (supple-
mentary text).
Thirty-six percent of the lethal events took
place within the local band; for example, be-
tween brothers, father and son, mother and child,
in-laws, husbands and wives, companions, friends,
clan “brothers,” neighbors, and so on (table S2).
Six percent of all incidents involved husbands
killing wives. In most lethal events (85%), the
killers and victims were members of the same
society. The remaining lethal aggression events
involved persons from outside the society, such
as shipwreck victims, colonists, missionaries, or
neighboring indigenous cultures. Almost all of
the killers were male, whether they acted alone
or with others. Females were the killers or co-
perpetrators in only 4% of the events.
The reasons for the lethal events varied.
Given that most lethal aggression involved one
killer and one victim, the large number of personal
motives for killing is not surprising (Table 1 and
0510
Observations
15 20 25
Yahgan (4)
Aweikoma (10)
Botocudos (3)
Paiutes (8)
Kaska (3)
Slave (1)
Northern Salteaux (2)
Micmac (0)
Montagnais (4)
Copper Inuit (15)
Ingalik (6)
Yukaghir (1)
Gilyak (2)
Aranda (5)
Tiwi (69)
Vedda (4)
Andamanese (2)
Semang (0)
Mbuti (0)
Hadza (5)
!Kung (4)
Lethal events with
more than one
perpetrator but only
one victim
Lethal events with
more than one
perpetrator and more
than one victim
Fig. 1. Lethal aggression events with multiple perpetrators in 21 MBFS. The distribution of all
lethal events that involved multiple perpetrators is shown, based on whether there was a single victim
(31 events) or multiple victims (29 events), out of a grand total of 148 lethal events. The figure reflects
how the Tiwi are more violent than the other MFBS. Nearly half of the sample societies (10 of 21) had no
lethal events perpetrated by two or more persons. Three societies had no lethal events at all. For each
society, the total number of lethal events is in parentheses.
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013
271
REPORTS

table S3). Specifically, many lethal disputes in-
volved two men competing over a particular wom-
an (sometimes the wife of one of them), revenge
homicide exacted by family members of a victim
( o ften ai med a t th e sp eci fic pe rson responsible for
the previous killing), and interpersonal quarrels
of various kinds; for instance, stealing of honey,
insults or taunting, incest, self-defense or the pro-
tecti on of a lov ed -o n e, an d so on. Additionally, in
some killing events, another person or persons
supported a companion who acted out of per-
sonal, not group, motives, such as when a friend
assisted a husb and in killing his wife’s lover (see
table S4, case 109).
About one third of the lethal events involved
disputes between people of different groups (T able
1). However, three-quarters (38 of 50) of in-
tergroup disputes took place among the Tiwi
alone. The percentage of intergroup disputes for
the T iwi exceeded 50% of their events, whereas
the corresponding percentage for the other twenty
societies was only about 15%. Another feature
of the Tiwi data is the regular occurrence of
strings of killings. Thirty-nine percent (27 of 69)
of Tiwi lethal events occurred in seven separate
strings (table S4, cases 129 to 131, 133 to 139,
140 to 141 and 145, 142 to 144, 146 to 149, 150
to 153, and 156 to 158), whereas only two strings
of killings occurred in the other 20 societies
(table S4, cases 10 to 13 and 82 to 89).
The findings suggest that MFBS are not
particularly warlike if the actual circumstances
of lethal aggression are examined. Fifty-five per-
cent of the lethal events involved a sole perpe-
trator killing only one individual (64% if the
atypical Tiwi are removed). One-person-killing-
one-person reflects homicide or manslaughter,
not coalitional killings or war . Additionally , 36%
of all lethal events occurred within the same
local group (62% if the atypical Tiwi are re-
moved), and violence within a local group is not
coalitional war . Only 15% of the lethal events
occurred across societal lines. Some such events
might fall within a definition of war , whereas
others might not (such as when shipwreck sur-
vivors were killed). Finally, very few lethal dis-
putes were over resources. Overall, a consideration
of reasons for lethal aggression reveals that most
cases stemmed from personal motives consistent
with homicide and, in some cases, family feuds,
but much less often with lethal aggression be-
tween political communities, or warfare (supple-
mentary text).
Approximately half of t he societies had no
lethal events that involved more than one per-
petrator. This observation is incongruent with
assertions by Bowles (9) and Pinker (14) that
war is prevalent in MFBS or by Wrangham
and Glowacki (8) that humans have an evolved
tendency to form coalitions to kill members of
neighboring groups. Additionally, two or more
persons killing a third person might or might not
occur in the context of a coalition against an-
other group or war. In some instances, motives
such as sexual jealousy (table S4, e.g., cases 29
and 109) or avenging the murder of a family
member (table S4, e.g., case 157) are clearly per-
son al ra th er than stemming from hypothesized
general hostility toward other groups. Most
mammalian aggression also is between individ-
uals. A different evolutionary perspective sup-
ported by comparative mammalian data, game
theory on the evolutionary logic of fighting,
and the observation that killing is an exceptional
event in human societies leads to the counter-
hypothesis that lethal behavior has been strongly
selected against, not favored, in comparison to
more restrained conflict behavior (7, 10, 15).
Taken together, the current findings contra-
dict recent assertions that MFBS regularly en-
gage in coalitionary war against other groups
(8), that “chronic raiding and feuding charac-
terize life in a state of nature” [(14), p. xxiv], or
that MFBS war deaths are substantial in recent
millennia and in the Pleistocene (9) (supplemen-
tary text). Perhaps discrepancies between the fore-
going propositions and the current findings can
be accounted for by proposing that self-selection
of ethnographic material may have exaggerated
war (10). Additionally, methodological factors
such as relying heavily on only a few secondary
sources (8), or estimating war mortality on the
basis of genocidal massacres and murders of
indigenous peoples by armed ranchers and
settlers (9), can lead to misimpressions in com-
parison to the use of systematic sampling pro-
cedures, reliance on primary ethnographic data,
and a focus on the specific circumstances of le-
thal aggression cases in MFBS (7, 10).
In conclusion, when all cases are examined
for a systematically drawn sample of MFBS,
most incidents of lethal aggression can aptly be
called homicides, a few others feud, and only a
minority warfare. The findings do not lend sup-
port to the coalitionary model. The predictions
are substantiated that MFBS, as a social type,
possess many features that make warfare unlike-
ly. The actual reasons for lethal aggression are
most often interpersonal, and consequently , the
particulars of most of the lethal events in these
societies do not conform to the usual concep-
tualization of war .
References and Notes
1. E. Leacock, Curr. Anthropol. 19, 247–275 (1978).
2. E. R. Service, The Hunters (Prentice-Hall, Englewood
Cliffs, NJ, 1966).
3. J. Steward, in Man the Hunter, R. Lee, I. DeVore, Eds.
(Aldine, Chicago, 1968), pp. 321–334.
4. R. Lee, R. Daly, in The Cambridge Encyclopedia of
Hunters and Gatherers, R. B. Lee, R. Daly, Eds.
(Cambridge Univ. Press, Cambridge, 1999), 1–19.
5. R. Tonkinson, in Keeping the Peace, G. Kemp, D. P. Fry,
Eds. (Routledge, New York, 2004), pp. 89–104.
6. R. Tonkinson, in War, Peace, and Human Nature:
Convergence of Evolutionary and Cultural Views,
D. P. Fry, Ed. (Oxford Univ. Press, New York, 2013),
pp. 262–277.
7. D. P. Fry, The Human Potential for Peace (Oxford Univ.
Press, New York, 2006).
8. R. W. Wrangham, L. Glowacki, Hum. Nat. 23,5–29
(2012).
9. S. Bowles, Science 324, 1293–1298 (2009).
10. D. P. Fry, in War, Peace, and Human Nature: Convergence
of Evolutionary and Cultural Views,D.P.Fry,Ed.(Oxford
Univ. Press, New York, 2013), pp. 1–21.
11. G. Murdock, Ethnology 6, 109–236 (1967).
12. D. White, Behav. Sci. Res. 23,1–145 (1989).
13. Materials and Methods are available as supplementary
materials on Science Online.
14. S. Pinker, The Better Angels of Our Nature (Viking,
New York, 2011).
Table 1. Reasons for lethal aggression, from the personal to the communal. The atypical
Tiwi findings are shown separately, followed by the other societies minus the Tiwi (n = 20), and
theentiresample(n = 21), all in number of cases (with percentages in parentheses). A more
detailed version of the table with case numbers for lethal aggression events is presented in
table S3.
Reason Tiwi only All others Total sample
Interpersonal events 24 (34.8%) 50 (63.3%) 74 (50.0%)
Revenge against a particular killer(s) 9 (13.0%) 8 (10.1%) 17 (11.5%)
Over a particular woman 8 (11.6%) 6 (7.6%) 14 (9.5%)
Over a particular man 0 (0.0%) 1 (1.3%) 1 (0.7%)
Husband kills wife 2 (2.9%) 7 (8.9%) 9 (6.1%)
Wife kills husband 0 (0.0%) 0 (0.0%) 0 (0.0%)
Miscellaneous interpersonal disputes* 5 (7.2%) 28 (35.4%) 33 (22.3%)
Interfamilial feud 0 (0.0%) 6 (7.6%) 6 (4.1%)
Within-group execution 0 (0.0%) 3 (3.8%) 3 (2.0%)
Execution of outsiders 4 (5.8%) 3 (3.8%) 7 (4.7%)
Intergroup events 38 (55.1%) 12 (15.2%) 50 (33.8%)
Over borders/resources (e.g., fruit tree) 0 (0.0%) 2 (2.5%) 2 (1.4%)
Theft of women from another group 0 (0.0%) 2 (2.5%) 2 (1.4%)
Interclan revenge-seeking 17 (24.6%) 0 (0.0%) 17 (11.5%)
During a general fight 4 (5.8%) 0 (0.0%) 4 (2.7%)
Miscellaneous intergroup disputes* 17 (24.6%) 8 (10.1%) 25 (16.9)
Accident 3 (4.3%) 3 (3.8%) 6 (4.1%)
Starvation cannibalism 0 (0.0%) 2 (2.5%) 2 (1.4%)
Grand total 69 (100%) 79 (100%) 148 (100%)
*For a listing of miscellaneous events, see table S3.
19 JULY 2013 VOL 341 SCIENCE www.sciencemag.org272
REPORTS

15. J. Maynard Smith, G. Price, Nature 246,15–18
(1973).
Acknowledgments: Some of the data reported here were
collected during research funded by the NSF (grant 03-13670).
We are grateful to the Svenska Kulturfonden for assisting
with the costs of publication. The data reported in this study
are located in the supplementary materials.
Supplementary Materials
www.sciencemag.org/cgi/content/full/341/6143/270/DC1
Material and Methods
Supplementary Text
Tables S1 to S4
References (16–127)
25 January 2013; accepted 10 June 2013
10.1126/science.1235675
Interactions of Multisensory
Components Perceptually Rescue
Túngara Frog Mating Signals
R. C. Taylor
1
and M. J. Ryan
2,3
*
Sexual signals are often complex and perceived by multiple senses. How animals i ntegrate signal
components across sensory modalities can influence signal evolution. Here we show that two
relatively unattractive signals that are perceived acoustically and visually can be combined in
a pattern to form a signal that is attractive to female túngara frogs. Such unanticipated
perceptual effects suggest that the evolution of complex signals can occur by alteration of the
relationships among already-existing traits.
H
uman perception of stimuli in multiple
sensory modalities can positively influ-
ence signal detection, selective attention,
learning, and memory (1). One example is “hear-
ing lips and seeing voices” in the McGurk effect
(2), which provided the foundation for speech
auditory-visual research (3). Studies of multi-
modal communication in animals have often
asked whether individual signal components in
different sensory modalities are redundant or car-
ry different information (4), but few studies have
investigated how specific interactions influence
signal perception (5).
Female túngara frogs base their mate choices
on male mating calls. Specifically, males produce
calls consisting of a whine alone or they may add
up to seven chucks; they do not produce only
chucks (6). Females exhibit phonotaxis (move-
ment toward a call, a bioassay of call recognition
and preference) to a whine only, but exhibit a
fivefold preference for calls with a whine-chuck
over a whine only [N =3662(11); see also Fig.
1A]. We tested female mate preferences in a
series of two-choice tests. Synthetic male vocal-
izations were broadcast from two speakers, one
of which was paired with a robotic frog that pro-
vided the visual stimulus of a calling male. Fe-
males were released equidistant from the two
speakers (with a 60° separation relative to the
female release point) and allowed to choose a
stimulus. Because our experimental configura-
tion differed from those of previous experiments,
we replicated some studies and obtained similar
results (Fig. 1, A, B, and D). Females were tested
only once.
The acoustic component of a frog’s mating
call is its most distinguishing feature, but visual
cues are also associated with the sexual display .
Male frogs have inflatable vocal sacs that shut-
tle air to and from the lungs while calling. Similar
to the movement of lips during human speech
(2), they are a biomechanical consequence of
the sound production system (7), but, as with
lips and speech, they can also influence the per-
ception of the ca ll (8, 9). W e have shown previ-
ously that female túngara frogs prefer a multimodal
signal (a call associated with a robotic frog) to a
call by itself (10), a result reconfirmed here
(Fig. 1D).
In túngara frogs, the temporal relationship
between acoustic components influences the
signal’s attractiveness (11 ). When the chuck in a
whine-chuck call is displaced by 500 ms, the call
becomes merely as attractive as a whine only
(Fig. 1B) and less attractive than a normal whine-
chuck (Fig. 1C). The temporal relationship be-
tween the acoustic and visual components of the
signal also influences the signal’s attractiveness
1
Department of Biology, Salisbury University, Salisbury, MD
21801, USA.
2
Section of Integrative Biology, University of Texas,
Austin, TX 78712, USA.
3
Smithsonian Tropical Research Insti-
tute, Post Office Box 0843-03092, Balboa, Ancón, Republic of
Panama.
*Corresponding author. E-mail: mryan@utexas.edu
Fig. 1. Preference responses. Each portion of the figure illustrates the acoustic components of the
túngara frog mating call: a whine only [(A, B,andJ), right gray], a chuck only [(J), left black], or a whine-
chuck (all other calls). The natural whine-chuck is depicted in (A), left black; (C), right gray; (D to G), all
acoustic signals; and (I), right gray. The rectangle represents the inflation-deflation cycle of the robofrog’s
vocal sac and its temporal relationship to the call [(D) to (J), left black]. The x axis represents 1000 ms,
green indicates the significantly preferred stimulus, and red indicates the unpreferred stimulus. In each of
the 10 experiments [(A) to (J)], 20 females were given a choice between the signal in black versus the
signal in gray. The vertical black and gray bars repre s e n t th e num b e r of fem a l e s that chose th e re s p e c t i v e
signal, and the blue dashed horizontal lines represent the null hypothesis of equal preference. Experiments
highlighted in the solid blue box are tests of the perceptual rescue versus template-matching hypotheses, and
those in the dashed blue box are the test of the component substitution hypothesis. The results of binomial tests
are noted as *** = P < 0.001, ** = P <0.01,*=P < 0.05, ns (not significant) = P > 0.05. The exact P values for
ea c h experiment are as follows: (A) P = 0.0003, (B) P =0.744,(C)P = 0.019, (D) P =0.034,(E)P = 0.323,
(F) P = 0.0049, (G) P =0.019,(H)P =0.039,(I)P =0.583,(J)P = 0.0001.
www.sciencemag.org SCIENCE VOL 341 19 JULY 2013
273
REPORTS