Male reproductive skew, females perspective and paternal birth cohorts

Introduction

Males and females have a conflict of interest, known as sexual conflict, because the investment in offspring differs considerable between the two sexes. Especially among mammals, a male is investing relatively less in each of his sperm compared to a female who produces relatively few large eggs of a limited number and combined with considerable costs during pregnancy and lactation. Males are expected to maximise their reproductive output by fertilising as many females as possible. On the other hand, females are expected to maximise their reproductive success by selecting mating partners of good quality such as good genes or qualities which help to rear their offspring successfully (Krebs & Davies 1993). Females are thought to be the limited resource that males compete for (Bateman 1948, Trivers 1972).

The concept of sexual selection goes back to Darwin (1871) who distinguished between the competition among males over access to oestrous females (male-male competition) and the choice of good-quality mating partners by females (female choice). Although there is this apparent conflict of interest between the sexes, males and females need to co-operate in order to reproduce. This compromise is expected to differ depending upon the mating system of a species.

The study species, the rhesus macaque, lives in groups consisting of many adult males and females who both mate with several partners. This results in a situation where all sexually mature males have to be treated as potential sires for any offspring considered. In species with multiple mating paternity can only be detected with the help of genetic analyses. It should be emphasised that paternity data give no information whether the outcome of reproduction is due to male monopolisation, female choice, both, or other factors. Paternity analyses do not uncover mating decisions; this can only be achieved by behavioural studies but mating strategies were not the aim of the present study. However, paternity analyses reveal kinship relations as a consequence of male and female mating decision.

This chapter aims to investigate kinship relations from three perspectives: the males’ perspective, the females’ perspective and the infants’ perspective. This chapter addresses the question whether the extent of paternal kinship in a social group of primates is pronounced or not (cf. J. Altmann 1979). I will present kinship data from six consecutive birth cohorts (1993-1998) in group R or its sister group BB (see Methods). For paternity analyses all sexual mature males were regarded as potential sires (see Methods), but in this chapter the number of potential sires will be restricted to the troop males, because including all males from the island will inflate the number of potential sires which addresses another question than asked here. Males from group BB are always treated as nontroop sires with respect to infants born in group R, but whether or not infants born into group BB are included in the analyses depends upon the specific question.

Males perspective

Studies across the animal kingdom have found evidence that males compete over access to fertile females using rituals or even overt aggression (Alcock 1984). With the use of paternity analyses it is now possible to demonstrate variation in actual male reproduction. In other words, not only studies investigating whether some males exclude others from mating can be conducted, also studies investigating whether males exclude others from reproducing can now be conducted. Assuming male-male competition, male reproduction is expected to be restricted to a few males which excluded other males from reproducing (hereafter: male reproductive skew). Again, whether the reproductive outcome is exclusively due to male monopolisation or whether female choice is influencing reproductive outcome also, will not be investigated here. The issue here is the degree to which true reproductive success among males is skewed.

Genetic studies of paternity have found that reproduction is restricted to a limited number of males each year, therefore male reproduction seems to be skewed in many primate species (e.g., Alouatta seniculus: Pope 1990, Macaca fascicularis: de Ruiter et al. 1992, Madrillus sphinx: Wickings et al. 1993, Papio cynocephalus: Altmann, et al. 1996, Presbytis entellus: Launhardt 1998, M. mulatta: Bercovitch et al. 2000, Semnopithecus entellus: Launhardt et al. 2001). Whereas some studies have shown that the highest ranking male sired more offspring than expected by an even distribution (de Ruiter et al. 1992, Launhardt 1998), others have found no evidence of rank influencing the reproductive success (Berard et al. 1993). However, the ability of a male to monopolise receptive females also depends upon the synchrony of female oestrous. If female synchronisation is strong, i.e., all females ovulate within a short time period, then a single male is less likely to be able to monopolise all fertile females. Especially in seasonal breeders, such as rhesus monkeys, where males need to mate guard in order to insure fertilisation, this time investment prevents males from mating elsewhere with other females, too. Here, the extent of skew in male reproduction will be investigated with other variables which may influences this skewness.

Females perspective

Contradictary evidence is known about the existence of female choice in primates (Huffman 1991, Manson 1992, 1995b) because it is difficult to separate the two confounding variables: is reproduction due to (i) female choice or (ii) male monopolisation or (iii) a combination of both or additional variables. A female needs to co-operate that the male can mount her, but harassment have been observed in some primate species (Neimeyer & Anderson 1983). As outlined above, investigating female choice is not the aim of the study, we here look only at the female perspective with respect to the kinship relations among her offspring. Are her offspring across years sired by different males or always by the same male? The latter case would increase the degree of relatedness among her offspring to full-siblings sharing on average a degree of relatedness of 0.5, whereas the former case would result in more genetic diversity among her offspring being half-siblings sharing on average a degree of relatedness of 0.25.

Infant perspective

When male reproductive success is strongly skewed within the mating season, infants born within the same year (hereafter: age-mates or peers) are likely to be paternal half-siblings (J. Altmann 1979). Because male rhesus only stay a few years in the group before they migrate again (Manson 1995a), paternal half-siblings within a social group are probably sired within a close time window. Depending upon the sharpness of this time window, paternal half-siblings are likely to be found in close age proximity.

Results Males reproductive skew

Male tenure and number of offspring sired

The mating season on Cayo Santiago usually starts in May and ends in October (see Methods). Census data from group R were available from the mating seasons of 1992-1997. All males greater than or equal to 5 years of age who were observed as group members in R for at least one of the six months of the mating season were counted as potential troop sires for the following birth cohort, i.e., a male present during the mating season of 1992 is a potential troop sire of any offspring born in 1993 regardless whether or not he was observed mating. Due to male inter-troop movement, group R had between 29 and 63 adult males during the of 1992 to 1997 mating seasons. Males who immigrated into group R during the mating season spend most of the whole mating season in the group (Table 3.1). Sires could also come from outside the group (see extra-group paternities below), but only troop males are considered in Table 3.1. Infants born in the sister group BB which fissioned during the mating season of 1996 (see Methods) are excluded here, so only infants born in group R and sired by potential troop sires were analysed.

All potential troop sires (5 year olds and older) were considered who spend 1, 2, 3 or all 6 months in group R during the mating season when infants were conceived, but born 6 months later (resulting in birth cohorts 1993-1998). Abbreviation are as follows: number of potential troop sires (N potential troop sire), *five potential troop sires could not been genotyped for unknown reasons and are therefore missing in the remaining analyses, number of infants with known paternity who were (i) sired by potential troop males and (ii) born in group R (N offspring considered), Spearman’s rank correlation coefficient (r_s) between the proportion of time spent in the group and the number of offspring sired is given for each year with the corresponding P-value.
Months

in group R

1

2

3

4

5

6

N potential

troop sire

N offspring considered

r _s tenure vs.
N off prod

P

1993

8

5

7

1

3

10

34 (30)*

30

0.357

P=0.053

1994

6

1

1

5

2

14

29

23

0.264

P=0.167

1995

13

9

2

5

10

24

63 (62)*

35

0.311

P=0.014

1996

10

3

4

12

4

21

54

30

0.049

P=0.726

1997

12

6

2

4

6

16

46

28

0.280

P=0.060

1998

7

1

13

3

4

17

45

30

0.242

P=0.109

Sum

56

25

29

30

29

102

271 (266)

176

Table 3.3.1: Male tenure and number of offspring sired

For three birth cohorts (1995, with a trend in 1993 and 1997) the data revealed a positive relationship between the proportion of time spent in the group during this mating season and the number of offspring sired. Manson (1995a) found in the same population as studied here, that male tenure is positively related with male rank, but the present study had just limited data on male rank (see male rank below). In some years (e.g., 1995) the number of potential troop sires was nearly twice as high as the number of infants born in group R indicating that some males will be excluded from reproducing because the number of males is higher than the number of offspring sired.

Actual sires

Most sires identified came from group R, but between 3.23 to 36.73 % of infants were sired from males outside of group R (hereafter: extra-group paternities), with most of them from neighbouring groups (see Table 3.2).

Note the total number of offspring is higher than in Table 3.1, because all infants (with solved paternity) born in group R or BB between 1993 and 1998 are included.
Cohort

N group paternities

N extra-group paternites

% extra-group paternities

1993

30

1

03.23 %

1994

23

13

36.11 %

1995

34

8

19.05%

1996

36

4

10.00 %

1997

31

18

36.73 %

1998

31

18

36.73 %

Sum

185

62

247

Mean

30.83

10.33

23.64 %

Table 3.3.2: Number of group paternities vs. extra-group paternities

The values of extra-group paternities were extremely high in birth cohort 1997 and 1998, where a high proportion of sires came from the sister group BB, which fissioned from group R in June 1996. In other words, even though these two groups completely fissioned and these males joined BB, they still sired offspring in group R during the mating season of 1996 and 1997.

Male reproductive skew among troop sires

If male reproduction is skewed the skew is either positive or negative. Positive skew describes a distribution, where few males have sired a higher number of offspring, but the majority of potential sires has no or few offspring. Negative skew, in contrast, describes a distribution, where few males have no or few offspring, but the majority of males has a higher number of offspring.

Two different skews among troop males were calculated. First, skew includes all potential troop sires, both sires and nonsires. Second, skew includes only troop sires, ignoring the number of nonsires. In the first skew calculation, nontroop sires and infants born in group R, but sired by nontroop sires, will be excluded, because a skew calculation using both troop and nontroop sires would have to include all potential troop and nontroop sires on the island which would inflate the number of potential sires to approximately 300-400 males. Therefore, the question is restricted to measure the degree of skew among troop males. Over 20 different ways to measure reproductive skew have been proposed (e.g., Cant 1998, Nonacs 2000, see review in Kokko et al. 1999), but the conditions and assumptions of the models do not fit most primates societies (cf. Clutton-Brock 1998). As a consequence, I used the standard measure of statistical skew (Sokal & Rohlf 1995) as a means of measuring the asymmetry in male reproductive success. The advantage of this measurement is that skew values can be compared across species. Table 3.3 gives a number of different variables which may have an effect on male reproductive skew.

Abbreviations are as follows: birth cohort considered (cohort), number of offspring sired by troop males and born in group R or BB (N off troop), number of potential troop sires (N pot troop sire), number of troop males who did not reproduce (N nonsire), number of different troop sires (N troop sire), number of different nontroop sires (N nontroop sire), percent of offspring sired by the top sire (% top sire), number of adult females in the group (N _{ad females} ), female cycle synchrony (Syn _i and Syn _ii in days, see below), statistical skew including troop sires and nonsires (Skew _i ) and statistical skew including only troop sires (Skew _ii ).
Cohort

N off troop

N pot troop sire

N troop nonsire

N troop sire

N non-troop sire

% top sire

N _{ad females}

Syn _i

Syn _ii

Skew _i

Skew _ii

1993

30

30

21

9

1

30.00

42

147

4.74

2.589

1.248

1994

23

29

20

9

13

21.74

46

112

3.03

1.968

1.578

1995

34

62

49

13

8

26.47

48

138

3.14

3.844

1.827

1996

36

54

38

16

4

27.18

56

125

3.05

4.168

2.586

1997

31

46

32

14

18

20.00

61

140

2.55

2.652

1.913

1998

31

45

31

14

18

19.35

55

223

4.65

2.499

1.984

Table 3.3.3: Male reproductive skew and variables which may effect male reproductive skew

As evident from Table 3.3, both measurements on male reproductive skew were positive, indicating that when sires and nonsires were considered the majority of males produced no or few offspring, whereas considering only sire, the majority of males produced one or few offspring. One the other hand, both skew calculations imply that few males also sired a higher number of offspring. The troop male which produced the highest number of offspring during a mating season (hereafter: top sire) was responsible for siring between 19.35 and 30.00% (mean ± SD=24.22 ± 4.44%) of infants born into the 1993-1998 birth cohorts. The remaining reproduction was shared by 8 to 15 other troop males. However, the number of nonsires varied between 20 to 51 males across years.

Male reproductive skew might be expected to be less pronounced when birth clustering is tighter, because more males have an opportunity to mate when more females are cycling at the same time. Assuming approximately similar gestation length, infants born around the same days are likely to be conceived around the same days which means that females were more or less synchronised in their ovulation. Male reproductive skew was not associated with female cycle synchrony calculated (i) as the interval between the first and the last birth of the cohort (r_s=-0.143, n=6, P=0.787) or (ii) as the mean interval over all consecutive births within a cohort (r_s=-0.086, n=6, P=0.872).

Three points should be mentioned with respect to other variables investigated even though their interpretation is limited because of the sample size. First, the higher the percentage of offspring produced by the top sire, the lower the number of nontroop sires (r_s=-0.943, N=6, P=0.005) as well as the number of offspring produced by nontroop sires (r_s=-0.986, N=6, P<0.001). Both associations may indicate that a top sire was not only excluding potential troop sires from reproduction (indicated by the percent of offspring produced by a single male), he also excluded potential nontroop sires.

Second, the more potential troop sires were in the group, the higher the number of offspring sired by troop males (r_s=0.928, N=6, P=0.008) which may suggest that the top sire was not be able to monopolise as many females as with less male-male competition, but also that less nontroop sires were successful in sneaky reproducing (cf. Berard at al. 1994).

Third, the higher the number of adult females in the group, the higher the number of troop sires (r_s=0.883, N=6, P=0.020). On the other hand, the number of adult females was neither associated with the number of nontroop sires (r_s=0.257, N=6, P=0.623) nor the percent of offspring sired by the top sire (r_s=-0.486, N=6, P=0.329). In other words, when more females were available they tended not to be fertilised by nontroop sires or by the top sire. Instead, the number of females was positively associated with the skew among sires (r_s=0.829, N=6, P=0.042), not with the skew calculated among both sires and nonsires (r_s=0.486, N=6, P=0.329) indicating that some troop sires probably had some success in reproducing few offspring each, when the number of females increased (shifting the positive skew curve a bit more to the right).

Top sires

The most reproductively successful male per season (top sire) differed across years. However, in 1994 two males (E05 and I37) sired the highest number of offspring, with E05 being also the top sire in 1993 (see Table 3.4). In addition, I78 and K85 were twice the top sire with a year in between where a different male was the most successful sire (see Table 3.4).

Values represent reproductive outcome of top sires including all infants born in group R, or its sister group BB between 1993 and 1998. The number of offspring sired by the top sire per season is marked in bold, with two top sires in 1994.
Cohort

1993

1994

1995

1996

1997

1998

sum

Top sire

N _offspring

N _offsprin _g

N _offspring

N _offspring

N _offspring

N _offspring

N _offspring

E05

9

5

0

0

0

0

14

I37

6

5

1

0

0

0

12

I78

0

0

9

8

9

0

26

K85

0

1

6

10

6

6

29

Table 3.3.4: Top sires and number of offspring they produced

What are the attributes of these top sires? Regarding rank, none of the top sires was the highest-ranking male of the troop at the time of reproduction. In fact, they were mid-ranking and rose in rank later in the season (Bercovitch unpubl. data, Kazem pers. comm., Widdig unpubl. data). Regarding age, the top sires varied in age between 8 and 11 years (mean ± SD=9.57 ± 1.27) at the time of reproduction, so they were in their “best” years with respect to body condition. Regarding natal status, all males were non-natal, i.e., none of the top sires was born in group R. Regarding kinship, the demographic database of the CPRC revealed, that only male I37 had a younger maternal half-brother (S19), but he did not join group R during the years investigated. Regarding group tenure, life history patterns of top sires seemed to differ from each other. For male E05, there was a peak in reproduction after which the number of offspring produced goes down, followed by a decreasing proportion of time spent in group R. In other words, he left the group as his reproductive success decreased. Similar, after his drop in reproduction, male I37 left the group but came back without being successful anymore in group R. In contrast, male I78 was never completely resident of group R for any mating season, but he sired a huge proportion of infants in the group. K85 was the most successful sire ever measured in group R with a total of 29 offspring produced and he was still in the group in 2000.

Male rank and age at the time of reproduction

No detailed data on male rank were available in the present study, but for the mating season of 1997, all potential troop sires (N=45) could been categorised by the outcome of dyadic agonistic interactions as either high-, middle- or low-ranking (each rank class including 15 males). High-ranking males sired 22 out of all 31 offspring (70.97%) born cohort 1998, middle-ranking males produced 8 out of 31 offspring (25.81%) whereas low-ranking males sired 1 out of 31 offspring (3.23%). In Table 3.5 a comparison was made between the number of sires and nonsires found in each rank class which revealed a significant association between sirehood and rank class (χ²=6.429, d.f.=2, P=0.040).

Number of sires and nonsires found in each rank class (high, middle and low). Data are restricted to mating season 1997 (and birth cohort 1998).
Rank class

High

middle

Low

Total

N sires

7

6

1

14

N nonsires

8

9

14

31

Total

15

15

15

45

Table 3.3.5: The association between sirehood and rank class

Testing the number of offspring as a function of the sires’ rank class revealed a marginal difference among the three rank classes (Kruskal-Wallis Test, χ²=5.836, d.f.=2, P=0.054). Berard at al. (1994) found in the same population that high-ranking males sired 55% of offspring by forming long-term consorts, while low-ranking males sired 45% of offspring by sneaking (i.e., quick and hidden) copulations. No clear advantage of one single mating tactic could be detected, but consortships might be most effective for high-ranking males (ibid.). Sneaking copulation is a reproductive tactic mainly used by younger males (Kuester & Paul 1992).

With respect to male age, results of the present study indicated that nontroop sires were indeed much younger than troop sires including only males older than 5 years of age (mean years ± SD: nontroop=10.13 ± 4.06, N=39 vs. troop sire=12.33 ± 4.26, N=73, Mann-Whitney U-test, z=-2.826, P=0.005). The extent of sneaky mating among troop sires cannot be evaluated, because of the lack of mating data. However, data suggest that male mating tactics were probably related to age. In contrast, age of troop sires did not differ from the age of troop nonsires (mean years ± SD: troop sires=12.33 ± 4.26, N=73 vs. troop nonsires=12.14 ± 4.54, N= 197, Mann-Whitney U-test, z=-0.359, P=0.720). Age-related male reproduction is illustrated in Fig. 3.1. Taking potential troop sires (5 years and older) between 1993 and 1998 into account, the graph represents the proportion between the number of infants and the number of males over age classes.

Fig. 3.3.1: Age-related male reproduction The graph represents the proportion between the number of infants and the number of males over age classes. With respect to the number of potential troop males per age class, only males between 9 and 11 years of age produced a higher number of infants than the number of potential troop sires available per age class. The same peak of reproduction was suggested by Bercovitch (1997) after analysing paternity in group R of one birth cohort. This result is also similar to that reported among male Barbary macaques (cf. Kuester et al. 1995).

Females Perspective
Continuity of paternity

From the females perspective it was measured whether females show continuity of paternity among their offspring. Given that a female gave birth to six offspring during the six cohorts investigated (1993-1998), paternity continuity would be high if a female conceived her offspring from the same sire each year. In contrast, paternity continuity would be low if she conceived her offspring from a different sire each year. By reproducing with the same male her offspring are full-siblings (degree of relatedness=0.5) whereas by reproducing always with a different male her offspring are maternal half-siblings (degree of relatedness=0.25). The proportion between the number of different fathers divided by the number of offspring would be one when the female always chose a different father (low paternity continuity) or the proportion would be close to zero when the female often or nearly always chose the same father again (high paternity continuity). As Fig. 3.2 illustrates, nearly 70% of mothers showed low paternity continuity as they always reproduced with a different father, the remaining percentage of females sometimes reproduced with the same father, but there was no single case where a female always reproduced with the same sire regarding all her offspring. This distribution differed significantly from a uniform distribution (Komogorov-Smirnov-test, z=4.992, N=52, P<0.001). Therefore, low paternity continuity seemed to be the general pattern for females of this group over the six years investigated.

Fig. 3.3.2: Continuity of paternity for reproducing females The number of mothers (N=52) in relation to the number of different sires per number of offspring. The graph shows that 36 out of 52 mothers (69%) reproduced with a different sire across consecutive years.

Number of full-siblings

Only 16 out of all 324 sibling dyads who shared the same mother and were born into group R or BB between 1993 to 1998 were full- not half-siblings. In other words, only 4.94% of all sibling dyads shared both the same mother and the same father. Fourteen of these 16 dyads were full-siblings born in consecutive years. There was only one case in which the same male sired the offspring of a female in three consecutive years. These results strongly suggest that (i) females reproduce with different males in consecutive years, therefore most individuals born to the same female will tend to be maternal half-siblings and (ii) if they are full-siblings (which is extremely rare), they show a tendency to be born in consecutive years, i.e., they are in close age proximity. Among the adult females investigated in the behavioural study (see chapter 4-6), there was none case of a full-sibling. Therefore, this kin category will not be included in the chapters that follow.

Number of maternal and paternal half-siblings vs. full-siblings

Including all infants born between 1993 and 1998 in R or BB with solved paternity (N=247) the number of maternal half-siblings, paternal half-siblings and full-siblings per individual was calculated over the six years. Results revealed that individuals had on average more paternal half-siblings (mean ± SD=9.887 ± 9.71, range=0-28) than maternal half-siblings (mean ± SD=2.623 ± 1.44, range=0-5) or full-siblings (mean ± SD=0.138 ± 0.38, range=0-2). Likewise the number of maternal half-siblings per individual was higher than the number of full-siblings (see Table 3.6).

Paired t-tests with respect to the number of paternal half-siblings (PS), maternal half-siblings (MS) and full-siblings (full-sibs) per individual born between 1993 to 1998 in group R with solved paternity (N=247).
Test

Paired t-test

PS vs. MS

t=11.718

N=247

P<0.001

PS vs. Full-sibs

t=15.972

N=247

P<0.001

MS vs. Full-sibs

t=-26.041

N=247

P<0.001

Table 3.3.6: Number of maternal and paternal half-siblings vs. full-siblings

This result was expected as the maximum number of maternal half-siblings and full-siblings in the six consecutive years is five, given that females bear only a single offspring per year (cf. Rawlins & Kessler 1986). In contrast, the maximum number of paternal half-siblings over six years can theoretically be much higher assuming male reproductive skew. However, it should be pointed out, that due to death and colony management the situation among the focal females in group R in 1997 was different (see below).
Paternal birth cohorts
The paternity data solved between 1993 to 1998 revealed that almost all individuals in the study group have paternal half-siblings because male reproductive success is strongly skewed. At least 73.96% of individuals had one paternal half-sibling of the same age. In addition, at least 15.02% of individuals had a paternal half-sibling within a two-year age difference. Hence, at most only 11.02% of individuals lacked a paternal half-sibling at all or in close age proximity within the troop (Table 3.7). These values were the lower limits of numbers or paternal half-siblings as paternity data of birth cohort 1999 and 2000 were not yet available. In sum, at least 88.98% of all individuals had (at least) a paternal half-sibling either of the same age or within a two year age difference.

Cell values show the number of infants who had paternal half-siblings (% are given in brackets). Abbreviations are as follows: number of individuals with at least one paternal peer (N pat peer), number of individuals with at least one paternal half-sibling born within two years of itself (N pat sib ≤ 2 years), number of individuals with either no paternal half-siblings or a paternal half-sibling at least three years older or younger than themselves (N no pat sib or ≥ 3 years), number of infants with solved paternity (Total = Solved). Note individuals counted in one category were not counted elsewhere, e.g., individuals with a paternal peer were not considered in the 2^nd category (paternal half-sibling born with two years of itself), therefore the number of individuals is equal to the number of infants with solved paternity.
Cohort

N pat peer

N pat sib ≤ 2 years

N no pat sib or ≥ 3 years

Total = Solved

1993

26 (83.87%)

4 (12.90%)

1 (3.23%)

31

1994

24 (66.67%)

7 (19.44%)

5 (13.89%)

36

1995

35 (83.33%)

3 (7.14%)

4 (9.52%)

42

1996

26 (65.00%)

8 (20.00%)

6 (15.00%)

40

1997

38 (77.55%)

7 (14.29%)

4 (8.16%)

49

1998

33 (67.35%)

8 (16.33%)

8 (16.33%)

49

Sum

182

37

28

247

Mean

73.96%

15.02%

11.02%

Table 3.3.7: Frequency of paternal sibship

These results demonstrate, that paternal sibships was very common in the study group during the six years investigated, because nearly all individuals had (i) at least one paternal half-sibling and (ii) paternal half-siblings were almost exclusive in close age proximity (see also Fig. 3.3)

Fig. 3.3.3 Percent of paternal half-siblings in close age proximity Solid black bars show individuals who have at least one paternal peer; striped bars show individuals who had at least one paternal half-sibling born within two years of itself, white bars show individuals who had either no paternal half-siblings or had paternal half-siblings at least three years older or younger than themselves.

These values represent the number of paternal half-siblings at birth, ignoring death and colony management. As a high proportion of 2-year-olds will be removed each year (see Methods), the number of paternal half-siblings born in same year was artificially reduced. As a consequence the number of paternal half-siblings per focal female used in the behavioural study was also reduced.

Discussion
The data presented in this chapter investigate kinship relations as a consequence of male and female reproductive decisions from three perspectives: the males, the females and the infants perspective. These three perspectives will be discussed in more detail below.

Reproductive skew

The most important result found from the male perspective can be summarised as follows. First, in the study group male reproduction among troop males is skewed within birth cohorts with few males producing a higher number of offspring whereas the majority of males produced no or few offspring. Male skew among potential troop sires was not associated with indirect measurements of female cycle synchrony, because skew was not decreasing when birth clustering is tighter, although more males should have an opportunity to mate when more females are cycling at the same time. Second, 24% of infants were sired by males from outside the troop (extra group paternities) meaning that these males were not excluded from reproduction by potential troop sires. Other studies found 36.4 % of extra group paternities in rhesus macaques on Cayo Santiago (Berard et al. 1993) and 33% of extra group paternities in Japanese macaques, M. fuscata, (Soltis et al. 2001), but paternity data were restricted to a small number of infants. The high proportion of extra group paternities might be a consequence of female choice as females seem to prefer novel males (Manson 1995a, Bercovitch 1991, 1997, Berard 1999) and nontroop sires were significantly younger than troop sires. In addition, nontroop sires tended to come from adjacent troops and when male rhesus macaques enter a new group they start at the bottom of hierarchy and move up only as higher-ranking males leave or die (Vessey & Meikle 1987). As nontroop sires occupy no dominance rank, these results indicate that factors in addition to male-male competition must influence male reproductive success. Third, on average 24% of offspring born per season were sired by the top sire (which was never the highest-ranking males of the group) whereas 20 to 51 troop males were excluded from reproducing at all. Fourth, limited rank data suggest that high-ranking males sired the majority of offspring born (71%) in one birth season whereas Berard at al. (1994) found in the same population high-ranking males siring 55% of offspring. Sixth, the peak of male reproduction was between 9-11 years comparing the number of infants with the number of potential sires per age class available which supports earlier findings on the same group after one year of paternity analysis (Bercovitch 1997). Studies on male Barbary macaques suggested a peak between 8.5-9.5 years of age (von Segesser et al. 1994) or found a positive correlation between male age and reproductive success (Paul & Kuester 1996).

The theory of reproductive skew was introduced by Vehrencamp (1983a,b) suggesting that reproduction is expected to be highly skewed when access to limited resources (e.g., fertile females) can be monopolised by only a subset of individuals. In other words, the dominant has complete control when settling alone or becoming a dominant is difficult for a subordinate (optimal skew models). Later versions of the model developed by Keller & Reeve (1994) added peace incentives (to forego challenging the dominant) and staying incentives (allowing subordinates to reproduce). Here reproductive skew is expected to increase with increasing relatedness between dominant and subordinate, as kin compensate loss in direct fitness with gains in indirect fitness (concession model). Most recent models on reproductive skew (Reeve et al. 1998) assume an influence on reproductive decisions by the subordinate suggesting an incomplete control by the dominant (incomplete control model). However, models of reproductive skew have mainly been developed to explain the variation in the distribution of reproductive opportunities among social insects (e.g., Keller & Nonacs 1993) and in co-operative breeders of birds or mammals (e.g., reviewed in Emlen 1997, Packer et al. 1991, Cooney & Bennett 2000, Allainé 2000). It seems unlikely that a single model will account for all eusocial invertebrates and co-operative vertebrates which show fundamental differences (Clutton-Brock 1998) and no available model applies to primate societies (cf. Emlen 1995, Sterck 2002).

There is evidence that competition between males primates over fertile females is strong and that there is a corresponding strong skew in the reproductive output (reviewed in van Hooff 2000). Skew in male reproduction has inferred in primate species based upon the finding that dominance rank is related to a larger proportion of mating opportunities (e.g., rhesus macaques: Chapais 1983, Hill 1987, Manson 1992, Savannah baboons: Hausfater 1975, Bulger 1993, see Cowlishaw & Dunbar 1991, 1992 for review), but no correlation between male rank and number of matings was found in other studies (e.g., rhesus macaques: McMillan 1989, Savannah baboons: Bercovitch 1986). This contradiction gave cause for a huge debate about the importance of male dominance rank and whether or not dominant males achieve indeed a higher reproductive output. However, more recent studies using paternity data could confirm that high-ranking males indeed monopolise the majority of mates as they produce more offspring than low-ranking males (e.g., rhesus macaques: Smith 1993, Berard et al. 1993, Bercovitch & Nürnberg 1996, long-tailed macaques: de Ruiter et al. 1992, de Ruiter & van Hooff 1993, Barbary macaques: Paul et al. 1993, stump-tailed macaques: Bauers & Hearn 1994, toque macaques: Keane et al. 1997, Japanese macaques: Soltis et al. 2001, Savannah baboons: Altmann et al. 1996, sooty mangabeys: Gust et al. 1998, Hanuman langurs: Launhardt 1998, red howler monkey: Pope 1990). Smith (1993), Berard (1999) and Alberts et al. (2002) showed by using long-term data that a positive correlation between male rank and number of mating opportunities occurs only in some years within a single population. The present study has only limited data on male rank, but they indicate that high-ranking males sired a higher proportion of offspring than both middle- and low-ranking males, although the alpha male in none of the six years investigated was the most successful sire (cf. Smith 1993 for captive rhesus macaques). Therefore, male dominance rank influenced reproduction, but not as a linear function.

To date, no quantitative study on male reproductive skew in primates has used paternity data and only two studies according to my knowledge have examined male reproductive skew in mammals using genetic data. First, reproduction among male African lions is more skewed with increasing coalition size and non-reproductive helpers, in these coalitions are often close relatives (Packer et al. 1991). As a second example, reproduction among male spotted hyenas (Crocuta crocuta) was also skewed as non-natal males over a 10 year study sired 97% of all offspring even though they are of lower rank than natal males, indicating female choice in this species, where females are dominant over males, has an impact on the degree of male reproductive skew (Engh et al. 2002). The present study is the first study in primates quantitatively investigating male reproductive skew measuring number of offspring sired by potential sires for six separated birth cohorts. The combination of both mating behaviour and paternity data would be necessary to answer the question whether skew in male reproduction is due to male monopolisation or female choice.

Female reproductive decisions

As females tend to conceive their offspring from different males in consecutive years, paternity continuity from the females’ perspective is low. An important consequence of this is that most siblings sharing the same mother will therefore be maternal half-siblings, not full-siblings, which are very rarely in the study group. This has also been found in paternity studies on other macaque species (Japanese macaques: Inoue et al. 1991, toque macaques: Keane et al. 1997). By reproducing with different males across years, females reduced the degree of relatedness among their offspring from r=0.5 (full-siblings) to r=0.25 (half-siblings). Female reproductive decisions might be explained in the sense of optimal out-breeding (Bateson 1982) accepting a lower degree of relatedness towards the benefit of genetic diversity among their offspring. Recent research on juvenile Atlantic salmon has reported that juvenile Atlantic salmon (Salmo salar) have a heterogenous advantage that outweights the benefits of kin-biased behaviour, both at the individual and population level (Griffiths & Armstrong 2001). The theories of kin selection and heterogeneus advantage predict diametrically opposite effects. In other words, the latter suggests that competition is low with high genetic diversity (i.e., among non-kin) and high with low genetic diversity (i.e., among kin). In contrast, Hamilton’s kin selection theory (1964) predicts less aggression among kin. The presented data raise questions about whether lineage specific mate choice characterises rhesus macaques (Silk & Boyd 1983, McMillan 1986), because females are not maximising the degree of relatedness among their offspring.

Paternal cohorts

This study revealed that from the infants perspective, at least 74% of individuals have one paternal half-sibling of the same age and at least 15% of individuals have a paternal half-sibling within a two-year age difference. Hence, at most only 11% of individuals either lacked a paternal half-sibling at all or in close age proximity within the troop. In fact, nearly 90% of infants will have at least a paternal half-sibling either of the same age or within two-years of age difference. In addition, the average number of paternal half-siblings per individual was much higher than the average number of maternal half-siblings. The presence of such a large number of paternal half-siblings which also tend to be in close age proximity, should have evolutionary consequences in terms of paternal kin discrimination (cf. J. Altmann 1979), since animals are expected to benefit supporting a paternal half-sibling to a similar extent as supporting a maternal half-sibling. A second implication of these results is, that even in large troops, such as group R, where many males compete over a limited number of oestrous females within a mating season, male reproductive skew creates cohorts of paternal half-siblings.

Summary
When access to limited resources (e.g., fertile females) is monopolised by only a subset of individuals, than conditions are created which are conducive to reproductive skew among potential sires. Paternity studies in primates have found that reproduction is restricted to a limited number of males each year, therefore male reproduction seems to be skew, but to date statistical analyses of skew in the actual reproductive success of males has not yet been published. The results of the present chapter suggest that reproduction among males was skewed over six years investigated, indicating that some males monopolise the high proportion of mates by excluding others from reproducing. The factors most likely to account for male reproductive skew in this study are female choice and male-male competition, because high-ranking males had a higher reproductive success, but low paternity continuity and a high proportion of nontroop sires may also suggest female choice. The extreme skew in male reproduction yielded a kinship structure where 74% of infants had a paternal half-sibling within the same cohort. Given that male reproductive success is probably skewed in other primate species, then J. Altmann’s (1979) suggestion that paternal sibships are important in primate societies, should be re-emphasised. One of the most important consequences of reproductive skew in primate societies is that many individuals will have more paternal, than maternal, half-siblings during their life time.