Introduction

Life-history traits are the products of evolution. Environmental variability and unpredictability have been considered the major selective forces for life-history traits (Pianka 1970; Schaffer 1974). A series of Ross's works (1988, 1992a, 1992b, 1998) indicated that, by excluding the influence of body size or by restricting analyses to closely related species, primates living in more unpredictable habitats have higher birth rates and an earlier age at first birth than do species living in more stable and predictable habitats.

As Ross (1998) herself noticed, however, such a relationship between environmental factors and life-history traits is based on the assumption that highly variable, unpredictable environments lead to high levels of mortality. Charnov (1991) has produced a theoretical model of mammal life-history evolution. A brief summary of this model by Ross and Jones (1999) is as follows: "The model explains life-history allometry by assuming that the age at maturity is determined by adult mortality rates, which are in turn determined by the environment. When adult mortality rates are high, animals are expected to mature rapidly, so as to minimize their chances of dying, and thus maximize their lifetime reproductive success. Charnov assumes that growth continues up to the age of maturity and then stops, with resources then being diverted to reproduction. Thus, the model predicts that animals that mature late will have a larger body size than those that mature early, and that this will influence birth rate. In a stable population, birth rate is balanced by mortality and hence birth rate will be expected to be positively correlated with juvenile mortality rates." Purvis and Harvey (1995) have tested his model. An analysis of life-table data from 64 mammalian species, ranging across nine orders, supported most of Charnov's (1991) predictions. Namely, both adult mortality and juvenile (between birth and first birth) mortality negatively correlated with age at first birth (measured from weaning) when body size is controlled. Both of these were positively correlated with the birth rate. Ross and Jones (1999) analyzed data on primates and obtained results similar to Purvis and Harvey (1995): primates with high rates of mortality reproduce at an early age and have high birth rates even when body size is controlled.

Rowell and Richards (1979) depicted patas monkeys (Erythrocebus patas) as representative of a fast-maturing, quick-breeding primate species selected for an unpredictable savanna environment. In spite of large body size, the median age at first birth of their captive patas was about 3 years old, earlier than that of their captive forest guenons (Cercopithecus mitis, C. ascanius, and C. neglectus). Moreover, their median interbirth interval was about 1 year, shorter than that of the captive forest guenons. Later, Chism et al. (1984) reported that all females of known age that matured during the course of their study experienced first birth at 3 years old in a wild population of East African patas monkeys (E. patas pyrrhonotus). However, interbirth intervals and mortality rates of adult and juveniles of wild patas monkeys have not yet been published.

Based on long-term, although intermittent, observations (2 years and 4 months of 14 years), we present data on birth seasonality, age at first birth, interbirth intervals, mortality rates, age at first emigration, and population change of a wild population of West African patas monkeys (E. patas patas), Cameroon. With this data, we discuss the validity of the suggestion that patas are fast-maturing and quick-breeding (Rowell and Richards 1979) with a presumed high mortality rate (Ross 1998).

Methods

Study site

Our study site was the Kala Maloue National Park in the Extreme North Province of Cameroon beside the Chari River (lat 12 °09′N, long 15 °53′E). It is roughly divided into three vegetation types: the riverine area along the Chari and its branch support woodlands dominated by Morelia senegalensis, Acacia sieberiana, Mitragyna inermis, and Crateva religiosa, and the area away from the rivers support grassland and open woodland dominated by Balanites aegyptiaca, Acacia seyal, and A. nilotica. Jackals (Canis aureus and Canis adustus), rare spotted hyenas (Crocuta crocuta), (Kavanagh 1978) and domestic dogs are the only predators of patas. The climate is characterized by two sharply defined seasons: a wet season from the latter half of May to September, and a dry season encompassing the rest of the year. The mean annual rainfall and temperature during 1985–1988 were 497 mm and 28.9 °C, respectively. Further details of location, climate, and vegetation are provided by Nakagawa (1989, 1991, 1999).

Study periods, study animals, and data collection

We have intermittently conducted field research on the patas monkeys 11 times from the time H.O. began research in 1984 (Ohsawa et al. 1993) until 1997. Each study period was 1–6 months long. The study periods and observers are shown in Fig. 1.

Fig. 1.
figure 1

Study periods I-XI and observers (shown by initials)

Full size image

Demographic events (birth, deaths, disappearances, immi-emigrations) were monitored in an intensively studied heterosexual group (the KK group). Size of the neighboring group (the BB) was also monitored. Individual adult females of the KK group were recognized using natural markings and characteristics. During period I in 1984, large amounts of millet and peanuts were given to the KK group to habituate them to observers. After period II in 1986, as a general rule, provisioning was limited to the onset of each study period and thereafter they could be observed at a distance of several meters without provisioning. Period VI was an exception because some of the KK group were captured and released in the course of collecting blood samples (Ohsawa et al. 1993).

The size of the KK group drastically decreased between 1984 and 1987, and continued to steadily increase until 1994, then decreased again in 1997 (see Results). Since we conducted research at least once a year, except for in 1990 during a group size increase between period III in 1988 and period X in 1994, we could obtain relatively reliable demographic data for these periods. Then, age at first birth, interbirth intervals, mortality rate of the females, and age at first emigration of the natal males were calculated on the basis of these data. However, the mortality rate during the group size increase period is an underestimate as a long-term average rate. Mortality rate of the females was recalculated on the basis of data from all the periods between 1984 and 1997.

Calculation of female mortality rate at each age

To increase the sample size and/or obtain the mortality rate of older females, we employed the following methods of calculation. Mortality rates at age x between 1988 and 1994 were calculated on the basis of data for survivals/deaths of cohorts born after (1988–x). Those born between 1984 and 1997 were calculated on the basis of data for survivals/deaths of cohorts born after (1984–x). The mortality rate (q x ) at age x was calculated by D x divided by N x , where D x (the number of deaths) equalled the number of individuals who died at age x, and N x (sample size) equalled the total number of individuals who survived throughout age x or died at age x. In addition, female mortality rates at 0–2 years old were calculated on the basis of combined data on females, males, and sex-unknown individuals since no significant sex differences in mortality rate were found at 0–2 years of age (see Results).

Because of the intermittent nature of the observations, we needed three assumptions to estimate the mortality (disappearance) rate at each age. The first assumption was the date of disappearance. Almost all disappearances (14 of 19 during 1988–1994, 78 of 89 during 1984–1997) occurred in the gaps between two study periods. When the date of disappearance could not be determined, it was estimated to be the midpoint between the last day of a study period and the date of confirmation of disappearance in the subsequent study period. The second assumption was the number of births. Our study periods did not cover the birth season in 1985–1987, 1990, 1992–1993, and 1995–1997. So, we did not know the precise number of births in these 9 years. As a result, we estimated the number of births on the basis of the number of adult females (namely, 3 years old or older) and the mean birth rate (0.75) (see Results) as follows: (1) when the estimated number of births, namely, the number of adult females multiplied by mean birth rate, was larger than the number of individuals confirmed in the subsequent study period, the former value was employed as the number of births; (2) when the former value was smaller than the latter, the latter value was employed. For example, in 1990, 6 (the number of adult females) multiplied by 0.75 (mean birth rate) equals 4.5. This value was smaller than 5 (the number of 1-year-old juveniles confirmed in the subsequent study period in 1991, i.e., period VI). Then, the number of births in 1990 was estimated at 5. In other words, no individuals born in 1990 seemed to have died before the onset of the subsequent study period. The third assumption was the date of birth. Date of birth of individuals born in the above years was estimated at 23 January from the mean birth date (see Results) in each year. For individuals born before 1984, when we started our research, date of birth was estimated at 23 January in the year in which we estimated they were born. As of 1984, ages of infants (0 years old) and juveniles (1 or 2 years old) were estimated from their body size, and those of adult females were estimated at 3 years old (median age at first birth, see Results) plus the age of her oldest offspring.

Calculation of mean annual juvenile mortality and mean adult mortality

Annual mortality rates were calculated from the life tables obtained. Crude adult mortality rate was calculated as the total number of deaths in adulthood divided by the total sample size in adulthood. Average instantaneous adult mortality rate (M) was calculated as loge(1–1/E α), where E α is the average life expectancy at first birth (Charnov 1991). Average instantaneous juvenile mortality (Z) was calculated with the equation e −Zα=S α, where α is age at first birth and S α is survivorship of females to first birth (Charnov 1991). Average juvenile mortality was calculated as 1−e −Z.

Results

Birth seasonality

Exact birth dates were obtained for all 33 births (including 2 stillbirths) that occurred during the study periods. All of these births occurred only between January and February (i.e., periods I, III, IV, VI, and X). Although seven infants had been born by 1 January 1994 (the first observation day in period X), their birth date could be estimated at the end of December 1993 based on their coat color. As one of them was still attached to the umbilical cord, his birth date was estimated at 31 December 1993. Judging from the mother's expanding belly, one infant seemed to have been born at the beginning of March 1991 (just after the end of observation day in period X). Judging from vaginal bleeding and shrinking of the mother's swollen belly, one stillbirth also seemed to have occurred on 7 February 1994 in period X although a dead infant could not be confirmed.

The total number of potential births (i.e., the number of 3-year-old or older females-year, see age at first birth and interbirth interval) was 56. Thus, 75.0% (42/56 in total; periods I: 14/22; III: 6/6; IV: 3/6; VI: 7/9; and X: 12/13) of all the adult females gave birth during the five relevant study periods, including just before or after these periods. On the other hand, no births occurred during periods II, V, VII–IX, and XI.

These data show that a clear birth season is detected in our patas and falls from the end of December until the middle of February, corresponding to the mid-dry season. The mean birth date was 23 January (Fig. 2). Almost all the births in a year occurred within 1 month of this period.

Fig. 2.
figure 2

Birth date during five study periods (I, III, IV, VI, X) (n=42), including estimated date of 8 births occurring just after and before the study. Mean date of births is marked with an arrow

Full size image

Age at first birth

Nine females experienced their first birth between 1988 and 1994, although one of them (Kz) was probably a stillbirth (see above). Of these, the exact birth dates of two mothers (Ks and Fi) and six infants (Ks, Kl, Fi, Ht, Og and Si's infants) were unknown because they had been born outside the study periods. However, birth year can be easily estimated on the basis of body size since the maximum gap between two study periods was about 15 months (period V–period VI) and all births occurred during a distinct and limited birth season. In this analysis, we assigned the mean birth date of 23 January to the estimated birth year. Although stillbirth and live birth resulting in death in the gaps between two study periods are overlooked by this method, primiparous females can be recognized by changes in nipple morphology. Thus, we estimated that eight out of nine females experienced first birth between the age of 34 and 38 months and the remaining one at 48 months (median=36.5 months; mean=37.1 months) (Table 1).

Table 1. Age at first birth of females in the KK group of patas during 1988–1994. Mean birth date (23 January) is used as estimated birth date when mothers' birth date (*) or first infants (τ) are unknown, because mothers and infants were born in the gaps between study periods
Full size table

Interbirth interval

Forty-five live births and 3 stillbirths were recorded between 1988 and 1994 including those that occurred in the gaps between study periods. Twelve out of 15 females gave birth more than once during these periods. Thirty-three interbirth intervals were recognized. Twenty five of 33 interbirth intervals were 10–13 months (about 1 year) and the remaining 8 were 23–24 months (about 2 years) (median=12.4 months; mean=14.4 months) (Fig. 3). To determine the effect of infant survival on the length of interbirth interval, we examined separately (1) intervals following infant deaths within a year after birth and (2) intervals following live birth resulting in survivals to the end of the first year of life. The median intervals from each data set are (1) 12.3 months (n=5, including one stillbirth) and (2) 12.4 months (n=28). The mean intervals are (1) 13.8 months and (2) 14.5 months. There was no significant difference in length between intervals following infant deaths within a year after birth and intervals following live birth resulting in survival to the end of the first year of life. (Mann-Whitney U test, U=64, P=0.76, two-tailed).

Fig. 3.
figure 3

Interbirth intervals during 1988–1994 (n=33). Solid blocks are intervals following live births resulting in survival to the end of the first year. Open blocks are intervals following infant deaths within a year after birth. Median is marked with an arrow

Full size image

Female mortality rate during group size increase periods between 1988 and 1994

Eighteen females, 20 males, and 10 sex-unknown individuals, including 3 stillbirths, were born during 1988–1994. There were no estimated births (see Methods). Of these, 4 females, 9 males, and 4 individuals of unknown sex disappeared by 12 February 1994 (the last day of period X). Among those who had been born during 1984–1987, only one individual (female) was confirmed in 1988 and did not disappear until 12 February 1994. Table 2 shows the disappearance rate at each age of individuals of known sex. No significant sex differences in disappearance rate were found in 0–2 and 4 year olds (Fisher's exact probability test, P=0.42 for 0 year olds; P=0.23 for 1 year olds; P=1 for 2 year olds; P=0.17 for 4 year olds). Significant sex difference was found at age 3 (P=0.02).

Table 2. Disappearance rate at each age of sex-known patas monkeys at Kala Maloue during group size increase period (1988–1994). The number of disappearances is the number of individuals who disappeared at each age. Sample size is the total number of individuals who were confirmed in the natal group throughout each age or disappeared at each age. The disappearance rate at each age was calculated as the number of disappearances divided by sample size
Full size table

Females between the age of 2 and 4 years old never disappeared from the natal KK group. Even when we consider the 7 females that were identified in the KK group in 1988 (period III), 5 of them were still confirmed in the group in 1994 (period X), and their estimated age ranged from 7 to 16 years old. Solitary females were never seen. Therefore, all the females are considered to remain in the natal group throughout their lives, and disappearances are attributed to death. In contrast, all of the 9 males disappeared by the age of 5 years. Solitary males and an all-male association were often seen, especially in the mating season (Ohsawa et al. 1993). Assuming mortality of 2- to 4-year-old male juveniles is negligible (as is the case with females), their disappearance is due to emigration from the natal group.

Table 3 shows the age at last observation (i.e., youngest estimate) and age at the date of confirmation of presumed emigration (i.e., oldest one) of 5 natal males. The youngest estimate of age at first emigration was around 30 months. So, first emigration from the natal group was thought to occur between 2.5 and 4.5 years of age.

Table 3. The age at first emigration of males from natal group
Full size table

The life table of females was drawn using mortality rates at 0–2 years of age; calculated on the basis of combined data of females, males, and individuals of unknown sex(Table 4). Average instantaneous juvenile mortality rate and average juvenile mortality rate were calculated at 14.0%, and 13.0%, respectively, on the basis of 65.8% of survivorship to first birth (i.e., 3 years old). This life table is incomplete since two died at 8–9 years of age, but 4 were still alive at the age of 10 or older among 7 females who had been born before 1987. Then, average instantaneous adult mortality rate could not be calculated because a lack of average life expectancy at first birth. Crude adult mortality rate was calculated at 4.3%.

Table 4. Life table of female patas monkeys at Kala Maloue during group size increase period (1988–1994) using mortality rates at 0–2 years of age on the basis of combined data on females, males, and sex-unknown individuals. The number of deaths, Dx, is the number of individuals who died at each age. Sample size, N x , is the total number of individuals who survived throughout each age or died at each age. The mortality rate, q x , at each age was calculated by D x divided by N x . Survival rate, p x , at each age was 1–q x . Survival to age x, l x , is the probability of surviving from birth to the beginning of age x (l 0 p 0 p 1....p x )
Full size table

Figure 4a shows female survival curves and age-specific mortality rate during 1988–1994. Mortality rate of infants and 1-year-old juveniles was about 20%. That of 2- to 15-year-olds was constant at zero except for individuals 8–9 yearsold.

Fig. 4.
figure 4

Female survival curves (solid lines) and mortality rate (dotted lines) on the basis of data (1) during group size increase period (1988–1994) and (2) during all the periods (1984–1997)

Full size image

Population changes between 1984 and 1997

Figure 5 shows annual changes in group size, the number of adult females of the KK and the BB group, and rainfall. Mortality rates shown here were calculated on the basis of data between 1988 and 1994. During this period, the size of the KK group continued to increase. On the other hand, the group size of the KK group dramatically decreased from 45 to 16 animals between 15 April 1984 (the last day of period I) and 11 July 1986 (the first day of period II). All of the 8 infants and 10 of 22 adult females died during this interval. The group size of the BB group, one of the neighboring groups, also rapidly decreased from 44 to 12 animals during this interval. Heavy drought occurred throughout the Sahel region in 1984 (Folland et al. 1986) and Kala Maloue was not an exception. Annual rainfall in 1984 was only 275 mm, lower than the mean (480 mm) minus SD (158 mm) (hereafter, defined as heavy drought) annual rainfall between 1984 and 1997. It was also lower than the mean annual rainfall (hereafter, defined as drought) in the years before and after 1984. The group size and the number of adult females of the KK group decreased again between 12 February 1994 and 20 July 1997 (from 13 to 9 in adult females; from 37 to 26 in group size) when drought or sub-drought continued over 3 years. Yet, heavy drought preceded by drought in 1989–1990 did not cause a decline in group size. Thus, a single year of heavy drought may not be a severe problem, but drought over 3 years is a problem for the patas monkeys at Kala Maloue.

Fig. 5.
figure 5

Annual changes in group size (solid lines), the number of adult females (dotted lines), and rainfall (annual: bars; mean annual between 1984–1997: horizontal dashed line). Circle represents the KK group; triangle represents the BB group. Although the influx of males into the heterosexual group temporally occurred during the mating season (Ohsawa et al. 1993), only one adult male remained outside the mating season. Then, the one adult male is counted in the group size. Rainfall data during 1983–1994 was recorded at Kousseri and during 1995–1997 at Ndjamena Airport, which are approximately 20 km from the encampment of Kala Maloue National Park

Full size image

Female mortality rate during all periods between 1984 and 1997

Twenty-three females, 26 males, and 32 individuals of unknown sex, including 3 stillbirths, were born between 1984 and 1997. Added to this, there were 44 estimated sex-unknown births (see Methods). Of these, 11 females, 20 males, and 69 individuals of unknown sex disappeared by 10 August 1997 (the last day of period XI). Among those who had been born before 1983, 27 females and 6 males disappeared.

The life table of the females was drawn using the mortality rate at 0–2 years old and calculated on the basis of combined data on females, males, and individuals of unknown sex(Table 5). The oldest female died at an estimated 17 years of age. Average instantaneous juvenile mortality rate and average juvenile mortality rate were calculated at 45.5% and 36.5%, respectively, on the basis of 25.6% of survivorship to first birth (i.e., 3 years old). Average instantaneous adult mortality rate was calculated at 26.8% on the basis of a 4.25-year life expectancy at first birth. Crude adult mortality rate was calculated at 21.6%.

Table 5. Life table of female patas monkeys at Kala Maloue during all of the periods (1984–1997), using mortality rates at 0–2 years old on the basis of combined data on females, males, and sex-unknown individuals
Full size table

Figure 4b shows female survival curves and age-specific mortality rate during 1984–1997, including drought over 3 years. Although age-specific mortality rates have large errors arising from the intermittent nature of the observations, one may sketch results as follows: Age-specific mortality rates during this period are about 20% higher than those during 1988–1994, mortality rates of infants and 1-year-old juveniles rose from 20% to 40%, and those of ages 2–13 rose from 0% to 20%. Mortality rate began to rapidly increase at about 14 years of age and reached 100% at about age 17.

Discussion

Birth seasonality and age at first emigration of male

The patas monkeys in Kala Maloue had a birth season restricted to about 2 months from the end of December to the middle of February within an 8-month-long dry season from October to May. In two populations of east subspecies and another population of west subspecies of patas, birth seasons are also restricted to about the same time of the year (November–February), which corresponds to the driest period within a 3- to 7-month-long dry season (Butynski 1988).

In general, a peak in birth is associated with the time of maximum food availability, which in the tropics occurs mostly during the wet season (Bronson 1985; Lindburg 1987; Rutberg 1987). Butynski (1988) concluded that this general rule could be applied to most guenon species, but not to patas monkeys. However, Nakagawa (2000) revealed that birth season (i.e., mid-dry season) for patas in Kala Maloue was associated with times when the monkeys could obtain more surplus energy and protein mainly from legumes of Acacia seyal and gums of A. sieberiana.

In our patas, emigration of a male from the natal group is considered to occur between the age of 2.5–4.5 years. This age at first emigration is also consistent with the typical age of 3 years in a wild Kenyan population (Chism et al. 1984). The Kenyan patas male thereafter lived alone or in shifting all-male associations until they reached adulthood at about the age of 5 years (Chism et al. 1984). Males aged less than 3 years were observed as neither solitary nor members in all-male associations at Kala Maloue.

Age at first birth and interbirth interval

Female life-history parameters of patas in Kala Maloue were compared with data of other patas and wild forest guenons (Table 6). In Kala Maloue, age (months) at first birth was 36.5 (median) and 37.1 (mean). Interbirth interval (months) was about 12 (median) and 14 (mean), independent of infant survival. These values are in agreement with data in a wild population of East-African subspecies of patas and/or captive patas. The interbirth interval is likely to be an overestimate since not only stillbirth but also live birth resulting in death in the gaps between study periods were overlooked. However, this does not seem to be the case because the mean interbirth interval (1.2 years) is shorter than the reciprocal of the birth rate (0.75) during the five study periods covering birth season.

Table 6. Comparison of life history parameters of patas with those of wild forest guenons
Full size table

Adult body weight and phylogeny influence the life-history parameters, and large primates are older at first birth and have longer interbirth intervals than do small ones (e.g., Harvey et al. 1987; Ross 1988). However, the largest-bodied patas have the youngest age at first birth and the shortest interbirth intervals than do their closely related smaller-bodied wild forest guenons.

Adult mortality and juvenile mortality

According to Charnov's (1991) theoretical model of mammal life-history evolution, adult mortality is considered a selective force and juvenile mortality as its final consequence: high adult mortality favors early and frequent reproduction. In a stable population, the birth rate must equal the mortality rate. The equality is insured by juvenile mortality being density dependent, and hence fecundity will be expected to be positively correlated with juvenile mortality (see Introduction for details). Data on mammals (Purvis and Harvey 1995) and on primates (Ross and Jones 1999) supports this model even when body weight is held constant. Those species with an early age at first birth and high birth rate for their body weight also show high adult mortality and juvenile mortality for their body weight.

Annual crude adult mortality rate of patas in Kala Maloue during the population increase period was 4%. When we recalculated the rate on the basis of life table during all the periods, including drought over 3 years, crude adult mortality rate rose to 22%. When the date of disappearance is estimated as the last day in a gap rather than the midpoint in a gap between study periods, the minimum estimate of mortality can be obtained. Even the minimum estimate of the crude adult mortality amounted to 18%. The crude adult mortality rate in wild forest guenons only come from C. mitis in Cape Vidal, South Africa (see Table 6). Despite their smaller body size than the patas, their adult mortality rate was 4%, which was the same and much lower than those of our patas during the population increase period and all the periods, respectively. On the other hand, the juvenile mortality rate during the population increase period is 13% (average) and 14% (average instantaneous). When we recalculated the rate on the basis of life table during all the periods, juvenile mortality rate rose to 37% (average) and 46% (average instantaneous). Even the minimum estimates (see above) of the juvenile mortality amounted to 31% (average) and 37% (average instantaneous). The juvenile mortality rate in wild forest guenons only come from C. ascanius in Kibale, Uganda (see Table 6). Despite their smaller body size, their average juvenile mortality rate was 10–12%, which was also a little and much lower than those of our patas during the population increase period and all the periods, respectively. These findings were consistent with Charnov's (1991) theoretical predictions in that patas monkeys with high adult and juvenile mortality showed early and frequent reproduction in spite of large body size, although more comparative and reliable data on mortality rate through long-term continuous observation is needed to be conclusive.

In Charnov's (1991) model, adult mortality is determined by the environment. In other words, adult mortality is determined by extrinsic factors (e.g., predation, parasitism, and climate), not by intrinsic factors (i.e., cost of reproduction). Almost all the deaths occurred in the gaps between study periods. We have no evidence as to the causes of death even for deaths occurring during study periods, except that a 6-month-old infant was killed by a jackal. We often observed the patas climbing up to the nearest tree and emitting alarm calls when they encountered jackals or domestic dogs. Given that almost all deaths were concentrated on 0- to 1 year-old individuals during the population increase period, however, adult mortality caused by predation is negligible during this period. Yet the adult mortality caused by climate seemed to be striking. Annual adult mortality rate amounted to 22–27% during all the periods, including drought over 3 years while it was 4% during the population increase period. Rapid decrease in the number of adult females was associated with unpredictable drought over 3 years. Nine out of 29 adult patas females died or disappeared during a period of severe drought in late 1980 and early 1981 in Laikipia, Kenya (Chism et al. 1984). Drought is likely to influence adult mortality through other agents, such as low food availability, parasites, and predators (Dunbar 1988; Milton 1996). At any rate, one can say that high adult mortality in patas is largely determined by the extrinsic factor of drought, and favors the evolution of early and frequent reproduction.

On the other hand, the predicted positive correlation between fecundity and juvenile mortality is derived from the assumption of a stable population (Charnov 1991). The patas population in Kala Maloue was not stable but composed of two phases: gradual increase and rapid decrease (see Fig. 5). Given that rapid decrease of population was associated with unpredictable drought over 3 years, high juvenile mortality in patas is not considered simply a density-dependent consequence of high fecundity.

Other evidence supporting the evolution of early and frequent reproduction

In many primate species, it is reported that the age at first birth and interbirth intervals in captivity or provisioned free-ranging populations became earlier and shorter, respectively, than those in the wild where nutritional conditions are generally poorer (Dunbar 1988; Lee and Bowman 1995). While patas monkeys are the exception (see Table 6), C. mitis, at least, are not the exception in that their interbirth intervals are shortened in captivity (Cords and Rowell 1987). Added to this, birth rate (8/9=88.8%) at primiparous age (3 year old) of the patas was rather high compared to the overall birth rate (75.0%) in contrast with many primate species (e.g., Macaca mulatta, Sade et al. 1976; M. sinica, Dittus 1975; M. fuscata, Sugiyama and Ohsawa 1982; Pan trogrodytes, Sugiyama 1994; Papio anubis, Strum and Western 1982). Sub-fecundity among primiparous females is partly due to the trade-off of individual growth for reproduction (Lee and Bowman 1995). The contradictory findings described above in patas female may suggest that they have an inclination to convert energy into reproduction, to some extent, at the expense of body maintenance or growth.

However, many lifehistory models assume that there is a trade-off between survival and reproduction (e.g., Charlesworth 1980; Charnov and Berrigan 1990; , but see Charnov 1991), that is, an increase in fitness due to earlier age at first birth and short interbirth interval is offset by a decrease in fitness due to lower chances of survival for the mother and offspring. While some single-species studies of primates supported this assumption, others did not (Ross and Jones 1999). Lee and Bowman (1995) proposed that the high neonatal mortality among primiparous females is partly a function of the trade-off between individual growth and energy diverted for reproduction. In our analysis on the basis of data during 1988–1994, there was no significant difference in the number of infants of primiparous (5 of 7) and multiparous females (25 of 30) which survived to 12 months of age (chi-square test, chi-square=0.47, P=0.52). Chism et al. (1984) also reported no significant difference in the survival rate of infants born to primipares and multipares in both captive and wild patas. These findings may suggest that energy conversion into reproduction from growth had only a negligible effect on the survival rate of offspring in patas, although its effect on the survival rate of the mother is unknown. Such traits also drive patas to early and frequent reproduction.