2 7
Langur Social Behavior and Infant Mortality
Richard A. Curtin
Behavior of the Indiclllangur monkey is highly
variable but follows two general patteirns. A pattern of
multi-male troops anld relatively low levels of aggres-
sion occurs in a wide range of habitats, hlut a patterin of
one-male troops and extremiiely high levels of aggres-
sion have been reportecl from crowded habitats and
those radically altered by humani influenice. High in-
fant mortality accomiipaniies the replacement of adult
males in troops within the second patterni, and this
mortality has been explained by a model of evolved
male reproductive strategy that includes infanticide.
The present paper reviews variabiility within the first
pattern and presenits a model in which the seconid
arises directly as the result of ecological pressure uponI
the first, aind infanit nmortality has no evolutionary sig-
nificance beyond the obviously detrimental. Research
strategies are proposed that will allow a choice between
the two models.
Presbytis entellus, the langur monkey of India, is now
the most thoroughly studied species in the colobine
subfamily. In part due to the colobine ability to digest
mature leaves, langurs can adapt to a wide range of
ecological conditions, and they live in most parts of the
Indian sub-continent except the western deserts. The
species has been studied at sites from Nepal to Sri
Lanka and in habitats ranging from Himalayan
meadow to central Indian forest, and it provides an
excellent opportunity to investigate the relationship
between differences in habitat and variation in social
behavior.
Langur social behavior is highly variable, but under
most conditions it follows a pattern in which troops
usually contain several adult males and levels of ag-
gression are relatively low. However, a distinctly diffe-
rent pattern was observed at the three sites of Dhar-
war, Jodhpur and Abu, India, where population den-
sities were high and ecological conditionis heavily in-
fluenced by main. Troops contaiined only single adult
males, and some of the most violent fighting ever
observed among monkeys occurred when males were
displaced from troops. L arge numbers of unweained
infants disappeared during these displacemenits, and
the explanation of their deaths is the subject of this
paper.
Hrdy (1974) explained the infant deaths in terms of
Trivers' (1972) discussion of sexual selection. The
deaths were attributed to male infanticide, and infan-
ticide was explained as behavior that evolved because it
acted directly to increase a male's reproductive success
at the expense of his competitors. Infanticide was an
evolutionary adaptation to conditions in which males
were frequently displaced from troops; when a male
established himself in a troop, he killed infants
fathered by the male previously in the troop so that
their mothers would more quickly become available
for reproduction.
This paper will question whether a phenomenon
that can lead to the loss of up to eighty-two percent of a
langur troop's infants over two birth seasons' is likely
to be the result of langur evolution. The occurrences at
Dharwar, Jodhpur and Abu will be viewed from a
perspective that emphasizes the uniqueness of both
the ecological conditions and the behavior at the three
sites, and a model will be presented that explains the
behavior as the result of ecological pressure on the
pattern of langur social organization typical of less
disturbed conditions. Infant mortality at the three sites
will be explained as being simply incidental to high
levels of aggression characteristic only of crowded
habitats and conditions far different from those in
which languir behavior evolved.
Variation in Langur Social Behavior
Nine studies of Presbytis entellus of at least four
months duration were carried out. The behavior ob-
served during these studies was highly variable, but
two basic patterns are apparent. The first was reported
from the wide range of ecological conditions at the
first six sites in Table 1. At these sites, several adult
male langurs typically lived in the same troop, and
males could move from troop to troop without major
social disruption. Non-troop males were few, and
there was no male interference with female reproduc-
tion. The second pattern has only been reported from
Dharwar, Jodhpur and Abu, sites either crowded or
radically transformed by man. Very high proportions
of troops contained only single adult males, and there
were numerous non-troop males. Males could appa-
rently enter troops only through a process of male
takeover in which males previously in the troops were
expelled, and the disappearance of unweaned infants
frequently accompanied these takeovers.
Differences in the behavior of adult male langurs
are fundamental to the two patterns. An inability of
one male langur to tolerate another in his reproduc-
tive troop appears basic to the second pattern, but
highly variable relations among males in the same
troop occur within the first. A careful consideration of
2 8
Table 1
SELECTED RESULTS OF STUDIES OFPRESBYTIS ENTELLUS
OF AT LEAST FOUR MONTHS DURATION
Habitat anid Human Inifluenice
Population Dentsity
Orcha, Madhya
Pradesh, India
Kaukori, Uttar
Pradesh
11/58-11/59    Jay 1962, 1965; Dolhinow  1972.
12/1/59-3/30/60
Deciduous forest, full complement of pre-
dators. Little human influence.
Rural cultivation, no predators.
10/26/62-5/11/63
Ripley 1965, 1967.
Partially cleared forest at archeological site sur-
rounded by cultivation, few predators. Some
human harassment of the monkeys.
Melemchi, Helambu,
Nepal
junbesi, Solu
Khumbu, Nepal
Simla, Himachal
Pradesh, India
11/71-9/72    Bishop 1975.
10/72-2/74    Curtin 1975; Boggess 1976.
8/72-1/73     Sugiyama 1976.
Oak-broadleaf forest with low level of human
use, probably full complement of predators.
Occasional human harassment.
Meadow and mixed forest, full complement of
predators. Occasional human harassment.
Oak - coniferous forest; human influence not
described but main study area the "western
suburb of Simla."
Dharwar, Mysore,
India.
10/61-3/63    Sugiyama 1964, 1965a, 1965b, 1966;
Sugiyama, Yoshiba and Parthasarathy
1965; Yoshiba 1967. Reviewed in
Sugiyama 1967, and Yoshiba 1968.
Dry forest, partially cleared, surrounding
forest recently cleared, few predators. Harass-
ment by domestic dogs.
134/km2 (forest)
jodhpur, Rajasthan
India
Abu, Rajasthan,
India
7/67-7/70    Mohnot 1971a, 1971b.
6/27/71 -9/13/71
6/19/72-9/12/72
2/11/73-3/26/73
Hrdy 1974.
Cultivation at edge of town and desert. Human
feeding and harassment, harassment by domes-
tic dogs.
Town and surrounding cultivation. Human
feeding and harassment, harassment by man
and domestic dogs.
*The population studied near Dharwar included troops living in roadside trees and
creek beds in cultivated areas; under these conditions, home range was estimated at 38
ha (30th troop). See Sugiyama, 1964.
the behavior at sites where it has been found that more
than one adult male langur can exist in a single troop is
necessary both to delineate the differences between
the two patterns and to illuminate the possible forces
involved in their origin.
Jay (1962,1965; Dolhinow 1972) described the rela-
tions among adult males as relaxed in both the virtu-
ally undisturbed forest at Orcha and the cultivated
fields at Kaukori. Superior observation conditions al-
lowed the detection of a stable dominance hierarchy
among the Kaukori troop's six males; a shift in rank
within this hierarchy was accompanied by fighting but
followed by a return to stability with both participants
remaining in the troop. Non-troop males were present
at both sites. At Orcha, a single male lived in the area of
overlap of two troop home ranges but was never seen
to interact with either troop, and at Kaukori, one of a
three member male group twice tried to join the study
troop but was rebuffed.
Sugiyama (1976) found that the langurs near Simla
followed a similar pattern. Most troops contained two
or more adult males, and relations among males were
usually calm with some evidence of stable dominance
hierarchies. Some males lived outside of troops for at
least the five months of the study, and their relatively
infrequent attempts at entering troops were actively
and successfully resisted by troop males.
Despite a much higher population density and fre-
quent, aggressive territorial behavior, Ripley (1965,
1967) found essentially the same situation at Polon-
naruwa. Troops contained two or more adult males in
stable dominance hierarchies, and although there was
some tension and a tendency towards spatial avoid-
ance among males, relations were usually calm. Male
groups of seven and eleven members were seen on
single occasions about six months apart, but no per-
manent non-troop males were known. Non-troop
males could enter troops at least on occasion, since two
sub-adult males successfully migrated between troops.
Two studies from Nepal both extended the area in
which multi-male troops were found and documented
the existence of powerful male rivalries within them.
Site
Dates
References
Polonnaruwa,
Sri Lanka
7/km2
7/km2
58/km2
15/km2 (min.)
1/km2
25/km2
not reported
50/km2 (min.)
-
Group Size
10-28
Home Range and Inter-troop Relations  Takeover?
Mean home range 380 ha; home range overlap,
avoidance between troops.
54     Home range 770 ha; only one troop at site.
19-c42  Estimated home range 32-128 ha; considerable
overlap with territorial defense by chasing
and display.
32     Home range 220 ha; considerable home range
overlap and possible inter-troop avoidance.
3-c19   Home range of main study troop 1200 ha; inter-
troop interactions rare because of range size.
16-98   Home range in Simla region 100 ha; little
inter-troop interaction.
9-24   Mean home range 16.8 ha;* territories with
mean size 8.9 ha are actively defended.
29-82   Not reported.
c20    Average home range 38 ha; inter-troop relations
not described.
The langurs that Bishop (1975) studied at Melemchi
showed a pronounced annual reproductive cycle, and
behavior among males changed with it. Sexual be-
havior was least frequent during winter, and interac-
tion among males was frequent and included groom-
ing, embracing and play. As the year progressed,
male-male grooming, then huddling and finally adult
male participation in play dropped from the reper-
toire of observed behavior. By the mating peak in late
summer, agonistic behavior among males had risen to
the highest levels observed, and male wounds were
most frequent. Summer also brought changes in the
troop's male membership; a large adult male left the
troop in late spring or early summer, and smaller ones
both left and entered in August. The great majority of
sightings of non-troop males also took place during
the summer, and these males were usually in the vicin-
ity of the troop rather than elsewhere as at other times
of the year.
The langurs that Curtin (1975) and Boggess (1976)
studied atJunbesi showed the same annual reproduc-
tive cycle, and male behavior again varied with the
cycle. Either an absence of interaction or low levels of
tension usually marked relations among males, but
agonistic interaction led on several occasions to males
being driven at least temporarily from the troop.
These peripheralizations were most frequent and pro-
found during the summer mating peak. All
peripheralizations during winter were temporary, but
two of the main study troop's four males were perma-
nently driven from the troop in May and a third was
periodically excluded until November, when the
troop's male membership stabilized at two. Behavior
consistent with this was observed in other troops. Both
fully isolated and peripheralized males were seen at all
times of the year, but most frequently in summer, and
non-troop males coordinated their movements with
nearby troops. One of the males driven from the main
study troop in May successfully entered a nearby
troop.
This pattern of increased agonistic behavior and
male peripheralization during the peak mating season
may have been present in the Melemchi forest as well.
Bishop's observations of changes in the male member-
ship of her study troop during the summer, and par-
ticularly of the increased number of isolates appa-
rently attracted to the troop then, are entirely consis-
tent with a model in which some adult males are at least
temporarily excluded from troops during the peak
mating season. It is possible that this pattern is quite
widespread, since both Jay's intensive study of social
behavior at Kaukori and Ripley's at Polonnaruwa were
carried out outside of peak mating seasons, and
Sugiyama's study included only the end of the peak
season at Simla.
The langurs discussed above share a social structure
in which most troops contain more than one adult
male but male rivalry exists. Rivalry varies in its ex-
pression from the relaxed dominance hierarchy at
Kaukori to the high levels of agonistic behavior during
peak mating seasons at Melemchi and Junbesi. Fight-
ing between males occurred even at Kaukori during
shifts in dominance rank, and there was an apparent
pattern of exclusion of males from troops during mat-
ing peaks atJunbesi and possibly Melemchi. Relations
between troop and non-troop males were usually tense
and accompanied by at least aggressive display, and
the success of non-troop males attempting to enter
troops was variable. Males were driven away from
troops at Kaukori and Simla, but entered troops at
Polonnaruwa, Melemchi and Junbesi.
Male rivalry within this pattern could cause changes
in troop membership, but there was no evidence that
these changes had any regular effect of reducing
numbers of troop males, and they took place without
any major disruption of troop structure. The tendency
for male exclusion from troops was balanced by the
observed ability of peripheralized males to successfully
enter their original or other troops, and so the number
2 9
.S
-
of males in any given troop could rise or fall. Exclu-
sions from troops were of long-term members in the
context of agonistic interaction with other long-term
members. Entry was a gradual process, and although
entries could follow prolonged agonistic behavior,
they were never linked with the exclusion of males
already in troops. The process of male takeover was
never observed, and changes in male membership
were never associated with infant injury or disappear-
ance.
Studies at Dharwar, Jodhpur and Abu revealed a
sharply contrasting pattern of langur behavior under
conditions of high population density and substantial
human interference with the animals. The situation at
Dharwar exemplified the remarkably similar social
behavior at the three sites. Most troops contained only
single adult males, and the rare multi-male troops
usually contained but one' large, fully adult male; al-
most half the adult males in the population lived out-
side of troops in male groups. Relations between these
male groups and the one-male troops formed the
heart of a pattern unique to these three sites among all
those where langurs have thus far been studied.
Male groups at Dharwar included individuals as
young as one year old and ranged in size up to thirty-
two members. They occupied relatively large areas
that overlapped the home ranges of several troops but
normally avoided contact with the troops. This pattern
of avoidance was frequently broken, particularly dur-
ing mating periods, by attacks by male groups on
troops. These attacks led to the most severe fighting
that has ever been observed among langurs, and they
were variable in their detail and their outcome. Any
number of the members of a male group might join
in an attack, and their incursion into a troop could
be met by female reactions ranging from flight to
approach and copulation. The male in the troop al-
ways resisted the encroachment of the male group and
could successfully drive off even a number of males,
but he might be defeated and himself expelled. Some-
times part of the troop left with the male group. When
a male group succeeded in expelling a troop's male, a
period of heightened agonistic behavior followed until
males of all post-weaning ages, whether originally
members of the troop or the invading male group,
were driven out and a single adult male remained with
the troop. Five of these takeovers were observed at
Dharwar, and the interval between their occurrence in
any given troop was estimated at three to five years.
Increased sexual activity and, frequently, attacks on
females with infants followed invasions by male
groups, and three takeovers were followed by the dis-
appearance of unweaned infants from the affected
troops. The investigators at Dharwar interpreted
these disappearances as being the results of attacks on
the infants by the males newly established in the troops
and noted that the mothers of the missing infants soon
became pregnant.
NIohinot (1971 a and b) described similar occu r-
rences atJodhpur. Onle-male troXps werIe againl typical,
and the number of non-troop males exceeded even
that at Dharwar. Mohnot witnessed the intrusion of a
male group on the surviving females and young of a
troop that had suffered mass mortality from unknown
causes. A single male drove out the others and re-
mained with the troop, and Mohnot saw this male kill
three infants and suspected that he was responsible for
the disappearance of two more. The Jodhpur sequ-
ence then departed from that observed at Dharwar in
that none of the mothers of the killed infants became
pregnant. The only female who conceived shortly
after the takeover had not been a mother, and hers was
one of the infants that disappeared.
Hrdy (1974) found the same pattern at Abu. She
repoirted the successive takeovers of two troops by the
same male, and then the return in the company of a
male group of the first male to be displaced. Un-
weaned infants disappeared during these takeovers,
and in several cases their mothers subsequently be-
came pregnant. Hrdy witnessed male attacks on
mother-infant pairs in one troop and received reports
from townspeople of males killing infants.
Two models of langur behavior
There are three major differences between the two
patterns discussed in the last section. In the first, most
troops contain more than one adult male and non-
troop males are few, males can move between troops
without major disruption, and infant mortality is in-
dependent of changes in the male membership of
troops. In the second, most troops have single adult
males and there are large numbers of non-troop
males, male entry into troops takes place within the
disruptive process of male takeover, and infants fre-
quently disappear during and after takeovers.
Neither geography nor subspecific differences sort
the two patterns. Populations following the first have
been reported from Nepal to Sri Lanka, and the full
expression of the male takeover/infant disappearance
pattern has been described from Mysore to Rajasthan;
both patterns occur within the subspecies Presbytis e.
entellus. The patterns are thus unlikely to be the results
of genetic differences between populations, and the
key to their differences must be sought in the details of
behavior and ecology. The successful model of langur
behavior must explain both.
Hrdy (1974) interpreted the behavior observed at
Dharwar, Jodhpur and Abu as being evidence of
highly regular patterns that were universal among
langurs. Her model followed that of the Dharwar in-
vestigators in its description of the history of a langur
troop as cyclic. Like Sugiyama, she believed that
fathers would tolerate their male offspring as adults,
and one-male troops would grow into multi-male
troops as infants matured within them (age-graded
troops as described by Eisenberg, Muckenhirn and
3 0
Rudran, 1972). The process of male takeover, how-
ever, was normal, and when it occurred, the growth of
a multi-male troop would be checked. A male group
would invade the troop and drive out all of its male
members, thus creating a new male group composed
of individuals of all ages, as observed at Dharwar and
Jodhpur. One male of the original male group would
then drive the others from the troop, and the cycle
would begin anew with a one-male troop. Differences
in the frequency of male takeover would explain varia-
tion in the proportion of one-male and multi-male
troops among different populations. At Dharwar,
where population density was high and there were
consequently large numbers of non-troop males,
takeovers were frequent and most troops did not have
time to mature into a multi-male configuration. In
areas like Orcha, where densities were low, takeovers
were so infrequent that troops usually contained two
generations of adult males and hence were multi-male.
The evolution of a pattern of male infanticide was a
major element of the model. Because they postponed
their mothers' fertility, unweaned infants were obsta-
cles to the reproductive success of a male newly enter-
ing a troop, and selective pressures favored male be-
havior that would remove the infants. These pressures
were strongest where population densities were high,
because frequent takeovers would shorten the average
tenure of males in reproductive troops. Since he would
probably soon be driven out, a male's promptness in
fathering offspring would be crucial. Because killing
infants quickly brings their mothers into estrus, infan-
ticide evolved essentially as an extension of male com-
petition for females into the past. An infanticidal male
enjoys accelerated and lengthened access to fertile
females at the expense of both his predecessor's and
the females' reproductive success.
The infant disappearances at Dharwar and Abu,
and the observed killings atJodhpur, were interpreted
as being the results of a male strategy of infanticide
that evolved in this manner. Hrdy rejected the possibil-
ity that the infant deaths were the results of aberrant
behavior and argued in favor of an evolutionary exp-
lanation of male attacks on the basis of the apparent
"goal-directed" and organized nature of those ob-
served (1974: 45), the apparent correlation between
attacks and non-paternity, and the existence of female
behaviors evidently evolved as "counter-strategies" to
male infanticide. An ability to evaluate the timing of
previous consort relationships was proposed as the
likely basis of a male's ability to avoid the evolutionarily
disastrous killing of his own offspring, and suggested
female counter-strategies included interference with
this mechanism as well as behavior more directly re-
lated to male attack. Pregnant females frequently con-
tinued to mate after takeovers occurred, and this
"pseudo-estrus" presumably could have resulted in a
male behaving towards as competitor's offspring as if it
were his own. Females sometimes cooperated in de-
fense of infants against male attack, and their behavior
during takeovers could include resistance of the at-
tacking male group and departure from the troop if
the takeover process was successful. At Dharwar,
mothers apparently abandoned their injured infants
and subsequently came into estrus, and Hrdy inter-
preted this as possibly reflecting accurate assessment
of the injured infants survival chances in the face of
continued attack. It was advantageous for the females
to concentrate their reproductive effort on future
offspring and to mate with the infanticidal males so
that their sons would possess the valuable genetic pre-
disposition towards infanticide (Hrdy 1974: 53).
Careful consideration of the variability of langur
behavior indicates the possibility of a different expla-
nation of the pattern observed at Dharwar, Jodhpur
and Abu. This second model emphasizes the crowded
ecological conditions at the three sites and states that a
conservative interpretation of the data available from
them indicates a substantially less regular pattern of
behavior than that described in the evolutionary
model discussed above. This less regular pattern is
explained as the result of the effect of crowding on the
widespread, ecologically adaptable pattern of langur
behavior discussed in the last section.
Under almost the entire range of conditions in
which langurs have been studied, troops were multi-
male and male rivalry was expressed solely by agonistic
behavior between males. This pattern was present in
the cultivated fields near Kaukori and the almost
wholly undisturbed deciduous forest at Orcha and was
observed in mixed forest and meadow at elevations in
excess of 3000m atJunbesi. It persisted under popula-
tion densities ranging from 1 langur per km2 at Jun-
besi to fifty times that in the relict forest surrounding
the archaeological site at Polonaruwa.
In contrast to this diversity, the pattern of one-male
troops, male takeover and high infant mortality has
been found only under crowded conditions or where
contact with humans was extensive. At the time of the
study there, the Dharwar area had been subject to
extensive recent deforestation and only a greatly di-
minished predator population remained; population
densities were as high as 134 langurs per km2 in what
forest survived. The Jodhpur and Abu langurs also
lived in habitats drastically altered by human influ-
ence. The Jodhpur langurs lived on the edge of the
desert surrounding the city, were extensively fed by
people and subsisted largely by raiding various types
of cultivation. The Abu population was even more
urban, consisting of "six troops... in and about the
town, and a seventh... at the Chippaberi bus stop,"
(Hrdy 1974: 21) and again fed by people and fre-
quently harassed. Such close association with human
habitation typically denies animals the use of consider-
able amounts of space, and so the effective population
densities at Jodhpur and Abu are greater than indi-
cated in Table 1.
3 1
The high population densities at the three sites are
probably of quite recent origin. They are almost cer-
tainly the results of the combination of habitat destruc-
tion by settlement, deforestation and cultivation and
the removal of the population check of predation.
Mukherjee (1974) states that habitat disruption and
faunal impoverishment became dramatic in India only
with the advent of the Mughal Empire and accelerated
with the coming of the British and again with Inde-
pendence. Schaller's (1967) work demonstrated that
both tigers and leopards frequently prey on langurs,
and the numbers of these predators has declined pre-
cipitously since the introduction of modern firearms.
Orcha, Melemchi and Junbesi, the three sites where
habitat destruction was least and predators were most
common, all had relatively low population densities.
Of all the sites where langurs have been studied,
conditions at Dharwar, Jodhpur and Abu were most
different from those in which langur behavior
evolved. The three sites are therefore the least likely
places to seek evolutionary interpretations of be-
havior, and data must be unequivocal before a wide-
spread pattern is inferred from occurrences observed
at these sites and nowhere else.
Table 2 summarizes the ten instances of male
takeover that are described in the current literature.
First-hand, eyewitness reports are available only of
those of the 30th and 2nd troops at Dharwar and the
B-26 troop atJodhpur, and of the 1972 takeover of the
B-6 troop at Abu. Regarding the actual causes of in-
fant disappearance, the genetic relationships of pur-
ported infanticidal males to missing infants and the
behavior of females within the takeover process, even
these accounts indicate a pattern that is much less
orderly than that described in the model of the evolu-
tion of infanticide.
The only direct observations of a langur male killing
infants have been those of Mohnot atJodhpur, and the
takeover that established the infanticidal male in the
B-26 troop followed the death of seventy-one mem-
bers, including the adult male, of an eighty-two
member one-male troop. Data were gathered at
Dharwar and Abu under less dramatic conditions but
did not include the results of any direct investigator
observation of infanticide. Behavior observed at all
three sites indicated that the takeover process might be
expected to lead to an increase in infant mortality even
in the absence of any male strategy of infant killing.
Infant langurs are quite vulnerable to injury from
attacks on their mothers as well as themselves, and
rapid arboreal locomotion exposes them to serious
risks of falling. Male takeovers are therefore likely to
be especially hazardous for infants; they involve the
most intense aggressive interactions ever reported
among langurs, and behavior includes extensive at-
tack, display and chasing among both males and
females. Sugiyama's (1965b: 387 et seq.) is the most
detailed account of the invasion and takeover of a
troop by a male group and the subsequent establish-
ment of a single male in the troop. Much of the be-
havior he describes creates risks to infants. Sexually
active females accompanied by infants initiated con-
tact with the male group; and male attacks followed.
Females followed the invading males intermittently
throughout the takeover process and were sporadi-
cally attacked. After a single male was established in
the troop, he attacked some mother-infant pairs, and
infants disappeared. In at least two instances, how-
Table 2
REPORTED CASES OF MALE TAKEOVER AMONG LANGURS*
Reference
Eyewitness?
Infants Missing    Evidence of Infanticide
Abu              2/73
B-6
*adaptedfrom Hrdy 1974:43
Sugiyama 1965b    Yes
Sugiyama 1966     Yes
Sugiyama 1967,
Yoshiba notes
..
Sugiyama 1967,
Kawamura notes
Mohnot 1971b
Hrdy 1974
6        Attacks on mother-infant pairs,
wounds found on infants.
4                  ,,
Not
reported
..
4-5       Not reported.
Yes
No
Yes
Yes, after
takeover.
5         3 cases witnessed; 2 inferred from disap-
pearances.
6         Disappearance, reports of 2 killings by
villagers.
4         Disappearance, report of 2 killings by
villager.
3         9 attacks on mother-infant pairs wit-
nessed; disappearance during observer's
absence from site; villager reports.
-         Attack on mother-infant pair witnessed.
3 2
Site,Troop
Date
6/62
7/62
7/62
10/62
3/63
7/69
7-8/71
Dharwar
30th
Dharwar
2nd
Dharwar
5th
Dharwar
Dharwar
Jodhpur
B-26
Abu
B-6
Abu
B-3
Abu
B-6
5/72
6-8/72
ever, his attacks continued after the disappearances
and it was apparent that the mothers, not the infants,
were the targets.
Mohnot's report confirms the occurrence of male
attacks on females, particularly when they attempt to
approach non-troop males. Such attempts do not ap-
pear infrequent during the takeover process, and it
seems that they lead to substantial threats to infant
safety. The dangers inherent in the high levels of
aggression that otherwise accompany takeovers are
more difficult to evaluate, but inter-male aggression
continued during the periods in which infants disap-
peared at all three sites from which disappearances
were reported. Two of the three killings observed by
Mohnot followed interactions between the infanticidal
male and members of the male troop of which he was
formerly a member. It is noteworthy that Mohnot's
interpretation of the only observed cases of langur
infanticide stressed the male's highly excited state
rather than any more ultimate cause.
The presence of more than one male at times of
infant disappearance obscures the identity of the male
involved in suspected instances of infanticide where
there has been no direct observation, but even if such
disappearances are assumed to be the results of infan-
ticide by males newly established in the infants' troops,
the genetic relationships between killer and killed re-
main unclear. Studies at both Abu and Junbesi indi-
cated that males frequently move in and out of the
same troop, and the possibility that infants disappear
during their fathers' return to their troops has yet to be
eliminated. Mohnot attributed the disappearance of
one infant at Jodhpur to infanticide by the male who
was in exclusive consort with its mother at the time of
its conception. This and the failure of males to kill
infants definitely fathered by rivals but born after
takeovers indicates that any correlation between infan-
ticide and non-paternity is incomplete. On the basis of
present data, the sounder correlation is between infant
disappearance and male takeover and requires no exp-
lanation in terms of an evolved male strategy of infan-
ticide.
Proposed female counter-strategies can also be
explained without reference to infanticide, and the
behavior of females within the takeover process ap-
pears to be much more variable than that described by
the infanticide model. The phenomenon of females
continuing to mate after conception has been observed
at both the Berkeley Animal Behavior Research Sta-
tion and at Junbesi and is a widespread aspect of lan-
gur reproductive behavior rather than any adaptation
to the specialized conditions at Dharwar, Jodhpur and
Abu. Cooperative defense of infants against male at-
tack is part of the pattern of cooperative caretaking
that is ubiquitous among langurs and survives even in
crowded conditions. Directly counter to predictions
in terms of female counter-strategies, females ap-
proached male groups at both Dharwar and Jodhpur,
and this behavior represented one possible way in
which females could contribute to the creation of situa-
tions in which infants were endangered. Data from the
Berkeley research facility indicate that the length of
time between birth and a mother's return to estrus is
highly variable among langurs and can be much shor-
ter than a year (P.C. Dolhinow, personal communica-
tion), and birth intervals of less than eighteen months
were observed at Junbesi. The possibility therefore
exists that mothers' resumption of estrous cycling is a
factor in their approaches towards male groups. If so,
the mothers' estrus precedes the loss of their infants,
and there is no need for infanticide to induce female
receptivity. Infant disappearances are again most
economically explained as results of heightened ag-
gression during the takeover process.
This description of male takeover as a chaotic, ir-
regular process in which infant langurs are en-
dangered indicates that the ultimate cause of infant
disappearance should be sought in the origin of the
pattern of one-male troops and male takeover. The
fact that this pattern is unique to crowded habitats is
evidence that its origin lies in the effects of crowding
on the widespread pattern of multi-male troops.
A review of Table 1 shows that small home ranges
are the most direct results of high population den-
sities. This is especially so in view of the fact that
effective home range is smaller in open or settled
habitat than in forest and that Dharwar should be
compared to Orcha, Melemchi and Polonnaruwa, and
Jodhpur and Abu to Kaukori and Junbesi. The effects
of small home ranges most likely to influence social
behavior probably lie in the proximity of neighboring
troops, the reduction of space between individuals
within troops and the limiting of areas available for
intra-troop social interaction.
The high levels of inter-troop aggression observed
at Dharwar and Polonnaruwa compared to Orcha and
Melemchi demonstrate the importance of the proxim-
ity of troops to the nature of inter-troop interaction,
but effects on troop structure are likely to be more
subtle. The langurs at Polonnaruwa maintained their
multi-male troops despite frequent territorial encoun-
ters, and it is probable that small distances between
troops affect troop structure through their influence
on intra-troop behavior rather than directly.
Small home ranges can increase levels of intra-troop
aggression by processes homologous to those well
known from studies of captive animals: reduction of
space between individuals and increases in competi-
tion over concentrated resources. None of the studies
thus far have reported details of intra-troop spacing,
but the observation of the Dharwar investigators that
troop spread in normal conditions ranged from 20 to
30 m and rarely exceeded 100 m can be compared to
the situation atJunbesi where the troop was frequently
spread over distances of 200 to 300m; there is at least a
possibility that crowding reduces intra-troop spacing.
3 3
The exact relatioinship betweeni cr-owdinig aind inltra-
troop aggression is similarly unclear, but the the ten-
sion and spatial avoidance among males at PolonI-
naruwa contrasts with the relaxed relations at Orcha
despite similar troop compositions at the two sites. Any
effect of crowding per se on aggression is probably
greatly exacerbated at both jodhpur and Abu by sub-
stantial levels of human feeding and harassment.
Crowding also reduces the amount of space availa-
ble for the resolution of aggressive interactions that do
occur. Here again, detailed studies have yet to be re-
ported, but agonistic behavior between males at sev-
eral sites took place across great distances. Males at
Dharwar were displaced up to 1 km during interac-
tions between troops and male groups; male groups at
Jodhpur have begun attacks on troops from 1.5 km
away; and a male at Abu was able to displace males
from the troop adjacent to his own. The peripheraliza-
tion and return of males to the Junbesi troop was
accompanied by active searching and threat at dis-
tances in excess of 1 km, and peripheralized males
elsewhere in the Solu coordinated their movements
with troops even more distant. These examples are
extreme, but they suggest the probablity that some
aggressive interactions between males cannot be re-
solved within the small home ranges typical of
crowded habitats.
The existence of multi-male troops at Dharwar,
Jodhpur and Abu would thus be difficult. Crowding
per se could be expected to exacerbate any chronic
inter-male tension similar to that observed at Polon-
naruwa and Junbesi, and heightened proximity would
increase competition over food, particularly where the
langurs were fed, and estrous females. Any serious
fighting could easily drive males beyond home range
or at least core area/territorial boundaries and into
contact with other, nearby troops. Crowding here re-
sembles the caged situation; the loser of an intense
aggressive interaction has no way to remain within the
social system. He is unlikely to be tolerated in his
precipitate entry into the range of another troop, and
so what might be a temporary exclusion in a less
crowded area will likely drive him permanently from
his troop. Processes such as that atJunbesi that involve
gradual reduction of great inter-male distances will be
interrupted by the intervening presence of other
troops. Crowding will thus both increase the fre-
quency of aggression and greatly magnify its effect;
loscrs become non-troop males.
One-male troops and male takeovers follow directly
from this creation of non-troop males. The wide-
spread pattern of increased aggression during mating
peaks, in combination with less systematic sources of
conflict, will regularly cause males to be driven from
troops. Since males apparently cannot enter troops in
crowded habitats, this will inexorably reduce the num-
bers of males in troops and finally bring about a one-
male troop pattern. The large numbers of non-troop
males will form male groups, and their proximity to
troops and the impossibility of their assimilation forms
the basis of the takeover pattern. Crowding causes
male takeovers, and the high infant mortality as-
sociated with them is an effect of crowding, not a result
of langur evolution.
Conclusion
The essence of the disagreement between the two
explanations of langur infant deaths discussed in the
last section lies in differing opinions of what is happen-
ing at Dharwar, Jodhpur and Abu. Such different
models arise from the possibility of different interpre-
tations of the data in three areas: the actual causes of
infant disappearances, the genetic relationships bet-
ween infants and purported infanticidal males, and
the relation between infant loss and female reproduc-
tion and behavior. Interpretation of the available data
as evidence of a highly regular pattern supports the
evolutionary model, but the same data can be taken as
indications of the chaos commonly expected in dis-
turbed ecology.
Conflict between the models will be resolved only
when our knowledge of langur behavior in crowded
habitats becomes more complete. Choice between
them awaits detailed understanding of both the pat-
tern of long term processes and the exact nature of
rare events, and the controversy over langurs
exemplifies the growing need in primate studies for
research strategies combining long term observation
and field and laboratory experiment.
Long term, continuous observation, preferably
conducted simultaneously by several investigators, is
needed both to understand processes that occur over
long time periods and to maximize opportunities of
seeing rare, naturally occurring events. Such study of
langurs will be necessary for an understanding of the
patterns of male movement among troops and male
groups, and in crowded habitats it will gain both the
crucial knowledge of the past relations with a troop of
the males involved in its takeover and the greatest
chance of seeing the takeover itself. Observation by
several investigators further increases the possibility of
witnessing a takeover and can provide histories of both
the troop and the male group involved.
Uninterrupted, long term observation is also neces-
sary to provide an accurate picture of female repro-
ductive cycles in habitats where male takeovers and
infant disappearances occur. Since birth intervals and
periods between birth and the resumption of estrus
vary widely from site to site and among captive lan-
gurs, an understanding of their pattern at the sites
where infant disappearance is studied is necessary for
a sound evaluation of the effects of disappearance on
female reproduction. The increased chance of direct
observation of takeovers will also help to illuminate the
pher ,nenon of female initiation of interaction with
male groups reported by Sugiyama and Mohnot.
3 4
Shorter or intermittent studies will be much less
likely to allow direct observation of takeovers and will
yield inadequate data on context. Neither will allow
the analysis of behavior in terms of annual reproduc-
tive cycles, which are known to have been important at
Dharwar, Melemchi and Junbesi. What behavior is
s5eeI CduL-illg short studies must usually be initerpl-etecd
on the basis of the assumption, shown to be incorrect at
both Abu and Junbesi, that males entering troops have
no histories of previous membership, and intermittent
studies are forced into the dangerous assumption of a
static situation during the observer's absence.
By itself, however, even long term study will proba-
bly prove inadequate to fully answer questions con-
cerning the takeover process itself and the immediate
causes of infant disappearance. Male takeovers and
infant disappearances are so rare and by their nature
difficult to observe that truly prohibitive investments
of observer time would be required to fully describe
them. Experiment is necessary both to define the con-
ditions that lead to takeovers and infant disappear-
ances and to create such conditions where the details
of the processes can best be observed.
The use of field experiement in primate studies is
not new. Kummer's (1971) classic studies of the basis of
differences in social structure between common and
hamadryas baboons made extensive use of release ex-
periments to demonstrate that the foundation of the
one-male troop pattern in hamadryas lay in male herd-
ing. Sugiyama (1966) attempted to examine the
takeover pattern at Dharwar by removing the adult
male from a one-male troop. The results of Sugiyama's
experiment illustrate both the problems and promise
of the experimental approach to the langur con-
troversy. The male's removal led to attempts, finally
successful, by males in adjacent troops to merge the
experimental troop with their own, but the normal
takeover pattern did not occur. Infant injury and dis-
appearnce followed the experimental removal, but the
observers were unable to witness the actual events.
Sugiyama's experiment did clearly demonstrate the
possible benefits of manipulation. The removal of
males from troops, combined with the release of diffe-
rent males, appears to present a powerful tool for
investigating takeovers and and their relation to infant
disappearance. This approach will allow direct inquiry
into the relation between crowding and the form of the
takeover process. If the experiment is performed in an
uncrowded habitat, Hrdy's hypothesis predicts that
infant disappearance will still occur. The alternative
model predicts that the results will resemble behavior
already observed at Junbesi: gradual male entry into
the troop unaccompanied by any injury to infants.
Experiment will also allow the manipulation of vari-
ables within the takeover process in crowded habitats.
Troops can be chosen for experiment on the basis of
from and released to the same troops after- varying
periods to test the correlation between infaint disap-
pear-anice and lack of geinetic relationship to mlales
enterinlg troops. The experimental process will also
greatly increase the quality of data that can be obtained
durinig a given takeover. The combiniation of removal
and release of males will allow the experimenter to
predict when aind where a takeover will occur, aind
observers and film equipment can be prepar-ed aild
present.
Colony studies are a further area of complementary
investigation. Introductions of males to troops con-
taining infants can be observed under optimal condi-
tions, and, uniqluely, the effects on female repro-
ductive cycles of male intr-oductions and infant loss caIn
be monitored physiologically as well as behaviorally.
The necessity of such extensive research is not uIn-
ique to the present controversy. The questions of
modern primate studies require both increasingly de-
tailed knowledge of what animals do over long time
periods and much clearer understanding of the causes
of behavior. The combination of long term observa-
tion with experiment is fundamental to the achieve-
ment of these goals, and it is crucial to the study of the
adaptive significance of social behavior if we recall
Williams' (1966: 261) statement in the midst of a diffe-
rent controversy:
"Parsimony demands that an effect be called a f'unc-
tion onily when chance can be ruled out as a possible
explanation. In an individual organism an ef'fect
should be assumed to be the result of physical laws
only, or perhaps the f'ortuitous effect of some unre-
lated adaptation, unless there is clear evidence that it is
produced by mechaniisms designed to produce it."
Interpretations of the evolutionary functions of social
behavior must be made on the basis of a sound know-
ledge of the context and the variability of the behavior,
particularly when evolutionary significance is sought
for only a small fraction of a large range of variation.
Variations in habitat - analogous to Williams' physical
laws - can be expected to lead to different expressions
of variable behaviors. Where ecological change is so
rapid that behaviors take place in contexts markedly
different from those in which they evolved, the dan-
gers of imputing evolutionary functions to their vary-
ing expressions - whatever their effects - should be
especially clear to anthropologists.
Notes
1Hardy (1974) attributed the deaths of nine of eleven
infanits borni to the B-6 troop at Abu, Raiasthani, over a
twenty-two month period to infaniticide.
the age of the animals they contain, single or multiple
adult males can be released, and males can be removed
3 5
REFERENCES CITED
Bishop, Naomi H. 1975 Social Behavior of Langur
Monkeys (Presbytis entellus) in a High Altitude Envi-
ronment. Ph.D. dissertation University of Califor-
nia, Berkeley.
Boggess, Jane E. 1976 The Social Behavior of the
Himalayan Langur (Presbytis entellus) in Eastern
Nepal. Ph.D. dissertation, University of California,
Berkeley.
Curtain, Richard A. 1975 Socioecology of the Com-
mon Langur, Presbytis entellus, in the Nepal
Himalaya. Ph.D. dissertation, University of
California, Berkeley.
Dolhinow, Phyllis J. 1972 The North Indian Langur.
In Primate Patterns. Phyllis Dolhinow, ed. Pp. 181-
239. New York: Holt, Rinehart and Winston.
Eisenberg, J., N. Muckenhirn and R. Rudran. 1972
The Relation Between Ecology and Social Structure
in Primates. Science 176: 863-874.
Hrdy, Sarah Blaffer. 1974 Male-male Competition
and Infanticide Among the Langurs (Presbytis entel-
lus) of Abu, Rajasthan. Folia Primatologica 22:
19-58.
Jay, Phyllis C. 1962 The Social Behavior of the Langur
Monkey. Ph.D. dissertation, University of Chicago.
Jay, Phyllis C. 1965 The Common Langur of North
India. In Primate Behavior: Field Studies of Mon-
keys and Apes. Irven DeVore, ed. Pp. 197-249. New
York: Holt, Rinehart and Winston.
Kummer, Hans. 1971 Primate Societies: Group
Techniques in Ecological Adaptation. Chicago: Al-
dine.
Mohnot, S.M. 1971a Some Aspects of Social Changes
and Infant-killing in the Hanuman Langur, Pres-
bytis entellus (Primates: Cercopithecidae) in Western
India. Mammalia 35: 175-198.
Mohnot, S.M. 1971b Ecology and Behavior of the
Hanuman Langur, Presbytis entellus (Primates: Cer-
copithecidae) Invading Fields, Gardents and Or-
chards Around Jodhpur, Western India. Tropical
Ecology 12: 237-249.
Mukherjee, A.K. 1974 Some Examples of Recent
Faunal Impoverishment and Regression. In Ecol-
ogy and Bogeoraphy in India. M.S. Mani, ed. The
Hague: Junk.
Ripley, Suzanne. 1965 The Ecology and Social Be-
havior of the Ceylon Gray Langur, Presbytis entellus
thersites. Ph.D. dissertation, University of California,
Berkeley.
Ripley, Suzanne. 1967 Intertroop Encounters Among
Ceylon Gray Langurs. In Social Communication
Among Primates. S. Altmann, ed. Pp. 237-254.
Chicago: Univ. Chicago Press.
Schaller, George B. 1967 The Deer and the Tiger: A
Study of Wildlife in India. Chicago: Univ. Chicago
Sugiyama, Yukimaru. 1964 Group Composition,
Population Density and Some Sociological Observa-
tions of Hanuman Langurs (Presbytis entellus). Pri-
mates 5: 7-37.
Sugiyama, Yukimaru. 1965a Behavioral Development
and Social Structure in Two Troops of Hanuman
Langurs (Presbytis entellus). Primates 6: 213-247.
Sugiyama, Yukimaru. 1965b On the Social Change of
Hanuman Langurs (Presbytis entellus) in Their
Natural Condition. Primates 6: 381-418.
Sugiyama, Yukimaru. 1966 An Artifical Social Change
in a Hanuman Langur Troop. Primates 7: 41-72.
Sugiyama, Yukimaru. 1967 Social Organization of
Hanuman Langurs. In Social Communication
Among Primates. S. Altmann, ed. Pp. 221-236
Chicago, Univ. Chicago Press.
Sugiyama, Yukimaru. 1976 Characteristics of the
Ecology of the Himalayan Langurs. J. Human Evol.
5: 249-277.
Sugiyama, Yukimaru, K. Yoshiba and M. Parth-
asarathy. 1965 Home Range, Mating Season, Male
Group, and Intertroop Relations in Hanuman
Langurs (Presbytis entellus). Primates 6: 73-106.
Williams, George C. 1966 Adaptation and Natural
Selection. Princeton: Princeton Univ. Press.
Yoshiba, K. 1967 An Ecological Study of Hanuman
Langurs (Presbytis entellus). Primates 8: 127-154.
Yoshiba, K. 1968 Local and Intertroop Variability in
Ecology and Social Behavior of Common Indian
Langurs. In Primates: Studies in Adaptation and
Variability. P.Jay, ed. Pp. 217-242. New York: Holt,
Rinehart and Winston.
Press.
3 6