The Influence of Dominance Rank on the Reproductive Success of Female Chimpanzees

Female chimpanzees often forage alone and do not display obvious linear dominance hierarchies; consequently, it has been suggested that dominance is not of great importance to them. However, with the use of data from a 35-year field study of chimpanzees, high-ranking females were shown to have significantly higher infant survival, faster maturing daughters, and more rapid production of young. Given the foraging behavior of chimpanzees, high rank probably influences reproductive success by helping females establish and maintain access to good foraging areas rather than by sparing them stress from aggression.

A. Pusey and J. Williams, Jane Goodall Institute's Center for Primate Studies, Department of Ecology, Evolution and Behavior, University of Minnesota,
1987 Upper Buford Circle, St. Paul, MN 55108, USA.
Jane Goodall, Jane Goodall Institute, Post Office Box 599, Ridgefield, CT 06877, USA.
*   To whom correspondence should be addressed. E-mail: pusey001@maroon.tc.umn.edu

Chimpanzees resemble these species in that they live in permanent social groups and feed predominantly on ripe fruit that often occurs in monopolizable patches (9). However, they differ in that females in some populations spend more than half their time feeding alone, and most females disperse to other groups before breeding, with the result that they are usually not surrounded by relatives (9). Compared to female macaques and baboons, it is difficult to detect linear hierarchies  among female chimpanzees. Although some female chimpanzees clearly dominate others (10-13), dominance behavior in stable groups or stable pairs of females is uncommon and is never observed between some dyads (10, 13). In addition, when aggressive behavior does occur within a dyad, it is sometimes two-sided with no clear winner (14). These observations have led some to believe that female dominance is unimportant for reproductive success (13, 15). However, others have suggested that dominant females may gain advantages (10, 16). Here we present data from a 35-year study of the chimpanzees of Gombe National Park,  Tanzania.

The chimpanzees of Gombe have been studied since 1960 (10). The 48.7-km2 park consists of a series of steep valleys running from the eastern rift escarpment (1600 m elevation) to Lake Tanganyika (775 m). The valley bottoms contain evergreen forest that gives way to semi-deciduous forest on the valley sides and grassland on the ridges (17). Since 1963, the chimpanzees of the central area of the park have been provisioned with bananas at an artificial feeding station in order to habituate them and to facilitate regular observation (10). The feeding station has been likened to an unusually long-term natural food source and, since 1970, has been estimated to provide less than 2% of the chimpanzees' diet (18, 19). Daily observations are made of the presence, reproductive state, and social interactions of individuals at the feeding station and, since 1975, during daily all-day follows of individuals throughout their range (10). Since 1970, the habituated community has consisted of 4 to 13 adult males, 10 to 18 adult females, and 18 to 31 immatures and has occupied a range of 6.75 to 14.5 km2 spanning three to six main valleys in the middle of the park (10, 20). Adult females spend about 65% of their time alone with only their dependent offspring, foraging in distinct but overlapping core  areas of about 2 km2 (19, 21), whereas adult males are more social, travel over the whole community range, and jointly patrol and defend its borders (10, 19). Whereas almost all males born in the community remain in the community as adults, most or all natal females visit other communities during adolescence, and about 50% emigrate permanently (22).

We assessed dominance relationships among females by examining the direction of all pant-grunts between females recorded from 1970 to 1992. Pant-grunts are the most reliable measure of submission in chimpanzees and correlate with the reception of aggressive behavior (13, 14). When we constructed dyadic matrices, any cells were empty, but by assessing 2-year blocks we were able to assign 88% of the females that were observed more than 10 days per year (23) as high-, middle-, or low-ranking in each block (24). Dominance rank was not related to body weight (R2 = 0.01, n = 15) (25, 26), but individual dominance rank increased with age (27) as observed in the chimpanzees of the Mahale Mountains (11). However, a female's rank at age 21 strongly predicts her rank a decade later (R2 = 0.80, P = 0.001, n = 9), suggesting that early rank acquisition is important.

Dominance rank has a marked effect on several measures of reproductive success. First, offspring survival was significantly related to mother's rank at the birth of her offspring. Infants of low-ranking females showed much higher mortality than those of high-ranking females over the first 7 years of life (Fig. 1). Second, the age at which daughters reached sexual maturity was significantly related to their mother's dominance rank (Fig. 2). Daughters of low-ranking females experienced their first full anogenital swelling during which adult males mated with them (28) as much as 4 years later than daughters of high-ranking females. Age at first full swelling was strongly correlated with age at first birth (R2 = 0.67, P = 0.01, n = 8 regularly observed females). Third, there was a tendency for high-ranking females to live longer (29). Finally, the annual production of offspring surviving to weaning age (5 years) was correlated with rank for mature females that were observed for at least  12 years (Fig. 3), indicating that low-ranking females were unable to compensate for the higher mortality of their offspring by reproducing more quickly. All these factors combined to produce higher lifetime reproductive success in females of higher rank (30). In the analyses concerning reproduction, we excluded the female GG, because she was sterile. GG was an aggressive, masculine-looking female who occupied the highest rank and cycled regularly for 28 years but never became pregnant. If she is included in the analyses, the relation of the rate of production of surviving infants and life-time reproductive success with rank ceases to be significant (Fig. 3) (30).

 
 Fig. 1. Kaplan-Meier cumulative survival plot of offspring at ages 0 to 7 for females of high, middle, and low rank at the birth of  the offspring. Dashed line indicates high rank (n = 10 infants of 4 mothers), dotted line, middle rank (n = 39 infants of 13  mothers), and solid line, low rank (n = 16 infants of 11 mothers). In a single Cox-proportional hazards regression with multiple  variables (38), mother's rank and age at birth were both significant factors. The exponentiated coefficients estimate the  proportional increase in infant mortality rates as a function of unit change in rank, such as from middle to low, or increase in age;  Exp(coeff.rank) = 4.3, P = 0.001, Exp(coeff.age) = 1.09, P = 0.01.
 
 

 
Fig. 2. Age of daughter's reproductive maturity plotted against mother's rank at the birth of her daughter (R2 = 0.59, P = 0.02, n = 9 females). For the one mother with two mature daughters, we used the mean age of maturity. Mother's average rank over the daughter's lifetime was also a significant indicator of daughter's age of maturity (R2 = 0.58, P = 0.02).

Fig. 3. Annual production of infants surviving to age 5 plotted against mother's average rank during her childbearing years (R2 = 0.36, P = 0.02, n = 14, excluding GG; R2 = 0.02, not significant, n = 15, including GG). Analysis includes all females whose rank was known during their childbearing years who were observed for at least 12 years as adults. 

Dominance probably exerts its effects in several ways. First, the high mortality of infants of low-ranking females in the first few months of life was partly due to the infanticidal behavior of the high-ranking female, PS, and her daughter, PM, who snatched and ate the infants of several females in the 1970s (10, 31). Since then, high-ranking females FF and GG were observed trying to snatch the newborn infant of middle-ranking female GM, and females of unknown rank in an adjacent community were seen eating another female's infant (20). These observations suggest that female infanticide may be a significant, if sporadic, threat, rather than the pathological behavior of one female. However, infants are vulnerable to infanticide for only a few weeks, and rank-related effects on offspring mortality continue well beyond that age (Fig. 1). Second, the younger age at which daughters of high-ranking females reach sexual maturity reflects their higher rates of weight gain (20), suggesting that high-ranking females have better nutrition. Better nutrition might also account for better survival of high-ranking females and their offspring. In species living in permanent groups, reduced reproductive success of low-ranking females has been attributed to chronic stress due to frequent aggression from other females (8). Because female chimpanzees spend so much time alone, often going for a day or longer without seeing another female, this is unlikely to be important in this species.

High rank may confer better access to food, both by enabling a female to acquire and maintain a core area of high quality and by affording her priority of access to food in overlap areas (32). Because of the mosaic distribution of vegetation at Gombe, some female core areas are likely to contain higher quality food than others. In addition, because core areas overlap almost completely, a high-ranking female may gain priority of access to preferred food sites in an overlap area. These modes of competition might explain why dominance behavior is less frequent in this species than in some others and why linear dominance hierarchies are hard to detect. If core areas are stable, competition is likely to be most intense when new or maturing females are attempting to establish their own core areas (16). This idea is consistent with the frequent observations of aggressive interactions from resident females to newly immigrant females at Gombe (33) and the Mahale mountains (11), the fact that PS and PM killed the infants of neighboring females as PM reached maturity (31, 34), and observations of more frequent dominance interactions during the establishment of a new female group in captivity (16). In addition, clear dominance relationships may only be established between females whose core areas overlap, thus explaining the general lack of clearly defined linear dominance hierarchies. The stability of core areas, the acquisition of core areas, and the relationships of females whose core areas overlap are the focus of current research at Gombe.

More research is needed to understand how female chimpanzees achieve high rank. Some females, such as GG, acquired high rank by their own aggressive behavior. Other females, such as PM, have gained high rank through their mother's support. The fact that alliances with kin are sometimes important makes it all the more striking that young female chimpanzees often disperse to other communities. This underscores the suggestion that they are "forced" to do so in order to avoid inbreeding because their male relatives do not disperse, perhaps because of even stronger advantages to males from cooperation with relatives (35, 36). More information on the relative importance of alliances in the acquisition of female dominance rank and the influence of dominance on reproductive success in other populations of chimpanzees might clarify why only half the females at Gombe disperse, whereas almost 100% do so in other populations (9). Finally, if the considerable degree of reproductive skew observed in the Gombe chimpanzees also occurs in other populations, this has implications for the future genetic diversity  of this endangered species. As populations become small and isolated, there is a greater chance for the genetic diversity of the population to be reduced by the successful reproduction of a few dominant individuals (37).

REFERENCES AND NOTES

   1.J. Walters and R. M. Seyfarth, in Primate Societies, B. B. Smuts, D. L. Cheney, R. M. Seyfarth, R. W. Wrangham, T. T. Struhsaker, Eds. (Univ. of
     Chicago Press, Chicago, 1987), p. 306.
   2.C. P. van Schaik, in Comparative Socioecology, V. Standen and R. Foley, Eds. (Blackwell Scientific, Oxford, 1989), p. 195.
   3.J. Silk, in (1), p. 318.
   4.L. G. Frank, K. Holekamp, L. Smale, in Serengeti II: Dynamics, Management, and Conservation of an Ecosystem, A. R. E. Sinclair and P. Arcese, Eds.
     (Univ. of Chicago Press, Chicago, 1995), p. 364.
   5.B. Chapais, in Coalitions and Alliances in Humans and Other Animals, A. H. Harcourt and F. B. M. de Waal, Eds. (Oxford Univ. Press, Oxford, 1992),
     p. 29.
   6.L. Ellis, Ethol. Sociobiol. 16, 357 (1995); A. H. Harcourt, J. Zool. London 213, 471 (1987).
   7.C. Packer, D. A. Collins, A. Sindimwo, J. Goodall, Nature 373, 60 (1995) [Medline].
   8.B. B. Smuts and N. Nicolson, Am. J. Primatol. 19, 229 (1989).
   9.T. Nishida and M. Hiraiwa-Hasegawa, in (1), p. 318.
  10.J. Goodall, The Chimpanzees of Gombe (Harvard Univ. Press, Cambridge, MA, 1986).
  11.T. Nishida, in Understanding Chimpanzees, P. G. Heltne and L. A. Marquardt, Eds. (Harvard Univ. Press, Cambridge, MA, 1990), p. 68.
  12.R. W. Wrangham, A. P. Clark, G. Isabirye-Basuta, in Human Origins, T. Nishida et al., Eds. (Univ. of Tokyo Press, Tokyo, 1992), p. 81.
  13.F. B. M. de Waal, Chimpanzee Politics (Harper and Row, New York, 1982).
  14.J. D. Bygott, in The Great Apes, D. A. Hamburg and E. R. McCown, Eds. (Benjamin/Cummings, Menlo Park, CA, 1979), p. 405.
  15.F. B. M. De Waal, Ethol. Sociobiol. 5, 239 (1984); R. W. Wrangham, Behaviour 75, 262 (1980) .
  16.K. C. Baker and B. B. Smuts, in Chimpanzee Cultures, R. W. Wrangham, W. C. McGrew, F. B. M. de Waal, P. G. Heltne, Eds. (Harvard Univ. Press,
     Cambridge, MA, 1994), p. 227.
  17.D. A. Collins and W. C. McGrew, J. Hum. Evol. 17, 553 (1988) ; T. H. Clutton-Brock and J. B. Gillett, Afr. J. Ecol. 17, 131 (1979).
  18.R. W. Wrangham, Anim. Behav. 22, 83 (1974) .
  19.___ and B. B. Smuts, J. Reprod. Fertil. Suppl. 28, 13 (1980) .
  20.Jane Goodall Institute's Center for Primate Studies at the University of Minnesota, unpublished data.
  21.Our unpublished data suggest that time spent alone is even higher, and 80% core area size smaller, than published figures.
  22.Of 14 females of known birth date born in the habituated community who experienced their first full anogenital swelling during the study, 6 settled as adults in
     the community, 5 transferred to other communities, and 3 disappeared. At the end of 1995, 5 of the 11 resident adult females in the community were born
     there, and 6 were immigrants.
  23.Each year 11 to 18 adult females were observed in the community, of which a median of 16 (range 11 to 17) were observed for more than 10 days. Of these,
     a median of 87.5% (range 38.5 to 100%) could be assigned ranks in each 2-year block.
  24.High-ranking females either gave no pant-grunts to any females or gave occasional pant-grunts to other high-ranking females, and received pant-grunts from
     middle- and low-ranking females. Middle-ranking females gave pant-grunts to high-ranking and some middle-ranking females and received them from
     low-ranking females and some other middle-ranking females. Low-ranking females rarely, if ever, received pant-grunts from any adult females but often gave
     them to middle- and high-ranking females. Fourteen females were assigned ranks of 1.5 or 2.5 during transitional periods.
  25.Individuals are weighed by luring them up a rope attached to a spring balance by placing a piece of banana in a tin attached to the top of the rope. We fitted
     the natural logarithm of all weights of nonpregnant females to a curve using LOWESS. For females carrying infants, we subtracted the weight of an average
     infant of that age. This method makes no assumptions about the underlying shape of the curve, so the curve reflects the patterns in the data as accurately as
     possible. This standardizes weights by age, so we can find the residuals for a specific individual and determine whether she is bigger or smaller than average.
     We then regressed the average residual for each individual against dominance and found no significant relation. Thus, high-ranking individuals are not larger or
     smaller than lower ranking chimps.
  26.L. E. Moses, L. C. Gale, J. Altmann, Am. J. Primatol. 28, 49 (1992).
  27.Eight females observed for at least two periods showed an increase in rank with age, one female showed a decrease, and 14 females showed no change
     (P = 0.04, n = 9, Sign test).
  28.A. E. Pusey, Behaviour 115, 203 (1990) ; J. Wallis, J. Reprod. Fertil. 109, 297 (1997) [Medline].
  29.Differences in survival are not statistically significant, but there is a suggestive trend with a Breslow-Gehan test (P = 0.07; X2 = 3.22) when females who reach
     age 20 are placed into two ranks at age 21, high (n = 9) and low (n = 6). This test compensates for the reduction in sample size at later ages by weighting
     earlier deaths more heavily. However, statistical power of the test remains low because n = 15.
  30.There were 11 deceased females, including the sterile female, GG, who were observed from or before the estimated age of 14 years (approximate age at first
     birth), whose rank and lifetime reproductive success could be determined. The relation between number of offspring surviving to 5 years and average rank is
     not significant when GG is included (R2 = 0.10, P = 0.34, n = 11) but is significant when she is excluded (R2 = 0.60, P < 0.01, n = 10).
  31.J. Goodall, Folia Primatol. 28, 259 (1977).
  32.One food source that needs to be taken into account at Gombe is the provisioned bananas, although they comprised only a small proportion of the diet. There
     was considerable variation in the number of days on which individual females received bananas per year (median = 17, range 0 to 107). However, there was
     no significant relation between the number of days on which individual females received bananas between 1970 and 1990 and their dominance rank
     (P = 0.62, n = 281 in a multiple regression taking individual into account; R2 = 0.02, P = 0.5, n = 12 when average number of banana days was regressed
     against average rank for each individual).
  33.A. E. Pusey, Anim. Behav. 28, 543 (1980) .
  34.___, ibid. 31, 363 (1983) .
  35.___, in (14), p. 465.
  36.___ and C. Packer, in (1), p. 250.
  37.The highest ranking female currently in the central community, FF, is 38 years old and has seven offspring, five over the age of 5 years. Two of her sons
     currently hold the alpha and beta position in the community, and one of her daughters resides in an adjacent community. She is the daughter of a high-ranking
     female (FLO) who produced at least 3 adult offspring, one of which was alpha male for 8 years. This family's genes are likely to gain significant representation
     in the total population in Gombe of less than 150 individuals.
  38.G. A. Fox, in Design and Analysis of Experiments, S. M. Scheiner and J. Gurevitch, Eds. (Chapman & Hall, New York, 1993), p. 253.
  39.We thank the government of Tanzania, Tanzania National Parks, and Serengeti Wildlife Research Institute for permission; A. Collins for ensuring the continuity
     of long-term data collection; numerous field assistants, students, and volunteers for data collection and analysis; T. Susman for organizing the demography
     database; and M. Tatar for assistance in performing the survival analyses. Many organizations have funded the project over the years, including the National
     Geographic Society, W. T. Grant Foundation, Wenner Gren Foundation, L. S. B. Leakey Foundation, Jane Goodall Institute, Carnegie Corporation, and
     NSF.

17 March 1997; accepted 10 June 1997
 

Volume 277, Number 5327, Issue of 8 August 1997, pp. 828-831

©1997 by The American Association for the Advancement of Science.

BEHAVIOR:

Enhanced: Subtle, Secret Female Chimpanzees

Around three decades ago, the myth of the passive female primate was at its height, bolstered by two assumptions. All females were assumed to be uniformly successful at reproduction, because "all females breed." And social evolution was thought to proceed mainly by sexual selection, in which female-female relationships played a small part. So when female cercopithecines--animals such as baboons and macaques--were found to have highly variable levels of success in breeding, correlated with an individual's life-long rank in the female hierarchy (her dominance status), the myth took a beating (1).[HN1] From then on, female strategies were seen to have their own logic independent of males, and the ecology of female primate relationships became a vigorous research area that was crucial to understanding social evolution. But there were a few holdouts. In some well-studied species--the great apes, for example--there seemed to be little variance in female reproduction. Selection on female reproductive strategies appeared particularly weak in these species, allowing females to be considered as relatively passive pawns of male maneuvering. [HN2], [HN3] A new study on page 828 of this issue suggests that this view was simply the result of our ignorance (2) and that dominant female chimpanzees are indeed more reproductively successful than their lower-ranked associates.

 Pusey, Williams, and Goodall (2) show that female chimpanzees in Gombe National Park, Tanzania, vary in their fitness and that fitness is correlated with dominance. [HN4] If it seems odd that it has taken 37 years for Jane Goodall's Gombe studies to yield data on female reproductive success, consider that the chimpanzee social community contains only 10 to 15 females at any one time, with a mean interbirth interval of 5 years between surviving offspring. [HN5],  HN6], [HN7] Half of the breeding females are immigrants, often of unknown age. Many females meet each other rarely, and even then clear-cut dominance interactions are infrequent, so that it can take several years for observers to detect the dominance relationship of any given pair. But by indexing the dominance ranking of the females over 2-year periods, Pusey et al. found that status correlated with at least three measures of fitness for breeding females. The effects are weak enough that if an infertile female is included in the data, the correlations disappear. Nevertheless the results gain strength from parallel correlations of female rank with infant production, infant survival, and offspring maturation rate. Indeed, the rate of production of weaned offspring by dominant females was almost twice as fast as by subordinate females, for whom it was 9 to 10 years.

 It is too early to say why rank has these effects, or why, if rank is as important as the new data suggest, females challenge each other so little. At Gombe, each mother's foraging is concentrated in her own core area within a larger community range defended by natal males. Escalated aggression among females is rare, but Pusey et al. speculate that such competition as there is, including infanticide by high-ranking females, may serve to acquire or retain the best core areas. What makes this reasonable is that in extreme circumstances female chimpanzees can compete as intensively as males. Thus captive females, meeting for the first time as adults, can use male-like behavioral strategies to gain rank, including opportunistic coalitions and frequent reconciliations (3). It may be that occasional female tactics, still seen too rarely for observers to assess their significance, stabilize ranks in adolescence and underwrite a lifetime of covert rivalry.
 

 

Better on top. Among female chimpanzees, as among males, rank in the dominance hierarchy determines reproductive success.

 Photo: R. Wrangham
 

Covert behavior by female chimpanzees may be important in the sexual realm as well. In the most substantial analysis yet of chimpanzee paternity, Gagneux et al. (4) genotyped 13 infants in Taï National Park, Ivory Coast, together with their mothers and all the males of their community. [HN8] Seven of the infants (54%) proved to have been fathered outside the community, a proportion far higher than anyone anticipated. In Taï, although females normally sleep together with the whole community, individual females sometimes disappear for days or weeks at a time. One of the cuckolding females was unseen by observers for only 1 day and another for 2 days during the months in which they conceived. Intense territorial competition means that male intrusions occur only in large subgroups, with fights but no matings, and are therefore rarely surreptitious. Gagneux et al. accordingly believe that extra-group paternity is a result of females visiting neighboring communities. The implication: Females are highly motivated to travel a long distance in search of the neighboring males.

Again, the small sample size implies caution. But even if the rate of extra-group paternity proves generally lower than 54%, a detectable rate of cuckoldry is remarkable and raises two significant points. First, female chimpanzees typically copulate several hundred times per baby, and only a small proportion of their copulations can occur with the neighbors. The data imply that females can select fertile times to cuckold their home-group males, suggesting they are choosing  genes.

Second, the costs to females of sneaky matings appear substantial, implying that they bring major benefits. Females risk severe aggression not only from the neighboring males, but also from males who are known to kill infants fathered from other communities. Yet it is hard to imagine how the genetic benefits can be very high, because females mate multiply. This means that even when a fecund female chooses to mate in a neighboring community, she could not, so far as we know, select a specific male as father. The phenotypic benefit thought to account best for multiple mating in chimpanzees is infanticide prevention, on the theory that males forgo aggression toward infants of mothers with whom they remember mating (5). In the Taï case, mating with the neighbors could provide an insurance  against future encounters with them when she is accompanied by her offspring.

But why should the female ever encounter the neighboring males in the future? Males have been dying at a high rate in the Taï study community. Could the cuckolding copulations be a preemptive safeguard against a take-over by neighbors? This might explain the curious observation that no neighboring females have been seen mating with the males of the study community. If so, the Gagneux et al. result may prove to be a rare phenomenon even among chimpanzees, specific to an unusual historical moment. Or is she insuring herself against a bad fruit-season in her own range, when she might need to eat in a neighboring range? That would suggest extra-group matings will be found commonly in other populations. Genetic data are being compiled quickly in other chimpanzee sites and should soon show whether high rates of extra-group paternity are common.

Until this year, no one suspected that female chimpanzees were so active in pursuit of their reproductive interests, yet they are probably doing still more than we appreciate. Chimpanzee and human social systems differ importantly in the characteristics of the female relationships, so there are certainly no direct analogies for human sociobiology. But these studies remind us that even where females interact rarely or subtly, female initiative can be a major force in the evolution of social systems. Selective impact doesn't necessarily correspond with social power.
 

References