Paternal kinship among adult female rhesus macaques (Macaca mulatta)Anja WiddigII1231.4 1.2.53.64.782.9 1.102.113.121314154.5.166.17187.19208.21229.23242627283.29Introduction301.2.313.Results1.32333435363738519#2042.39488#202403.4142430#261Discussion43444546Summary4.47Introduction1.482.3.49504.5.51Results1.5354552.5657583.5960614.63Discussion646566676869707172Summary 735.74Introduction75 Results761.772.783.794.80614#43681612#220Discussion828384 Summary856.86Introduction878889Results1.902.9192933.95964.9798991005.101380#219102380#219380#219103Discussion 104105106107108380#243109Summary1107.111 1.1122.1133.1144.115116 References117118119120121122123124125126127128129130131132133134135 Acknowledgements136137138 Publications142143144 Appendix 1: Study group 145577#887146 Appendix 2: Ethogram147148149150 Appendix 3: Paternity results151152 Erklärung153enTable of contentsHelp Paternal kin discrimination and age proximity
Introduction Mechanisms of kin discrimination

Kin selection is expected to promote the evolution of social behaviour that increases the survival and reproductive success of close relatives (Hamilton 1964, Maynard Smith 1964). One prerequisite for kin selection is that individuals have the ability to discriminate between kin and non-kin (Grafen 1990). There are four possible mechanisms underlying kin recognition (Box 4.1).

Box 4.1: Mechanisms of kin recognition.

Note, recent work suggested to drop spatial distribution and recognition alleles as mechanisms of kin recognition and combined familiarity via association and phenotype matching as one mechanism as both are based on learning (Tang-Martinez 2001).

Spatial distribution: If relatives are distributed predictably in space, nepotism (favouring kin) occurs as a result of location-specific behaviour, e.g., in reed warbler (cit. Krebs & Davies 1993).

Familiarity via prior association: Relatives must predictably interact in unambiguous social circumstances, in which kinship is rarely confused due to the mixing of unequally related individuals (social learning), e.g., in house mice (Kareem & Barnard 1982).

Phenotype matching: An individual learns its own phenotypic attributes (odour, appearance, vocalisations etc.) or those of known relatives and matches this learned template against a potential kin in order to assess kinship (recognition in the absence of familiarity), e.g., in ground squirrels (Holmes 1986a).

Recognition alleles or green beard effect: Genes that cause a unique phenotypic effect (e.g., green beard) enable the bearer to recognise them in other individuals, one case reported in the fire ant (Keller & Ross 1998).

Among mammals, maternal kinship is detectable for human observers as lactation reveals mother-infant relationships. These close mother-infant bonds produce a social system in which maternal kinship and familiarity are tightly associated. Familiarity is therefore thought to be the most important mechanism of maternal kin recognition in primates (Gouzoules & Gouzoules 1987, Walters 1987, Bernstein 1991, Chapais et al. 2001). However, maternal kinship frequently coincides with patterns of spatial proximity and a higher frequency of grooming and agonistic aiding among female primates, making it difficult to distinguish the effects of kinship from those of familiarity on structuring social relationships. Therefore it is also unclear, whether primates recognise indeed shared genes in terms of the actual degree of relatedness or whether they just recognise the degree of familiarity.

In multi-male societies with unknown paternity, where rank, age or other factors may influence male reproduction, paternal sibship is expected to be found (e.g., as a result of male reproductive skew, see chapter 3). Paternal kin are probably not familiar to each other as fathers are not likely to mediate familiarity among their offspring. Among primates it remains questionable whether paternal kin can be identified in the absence of familiarity or whether they use alternative mechanisms such as phenotype matching to recognise each other.

To be more precise, kin discrimination among primates is most likely to arise if individuals either classify relatives due to shared family traits, via phenotype matching (Holmes & Sherman 1983, Lacy & Sherman 1983) or identify relatives due to frequent association patterns, via familiarity (Walters 1987, Tang-Halpin 1991).

Defining kin discrimination vs. kin recognition

It is necessary to define relevant terms used in this chapter, as they have often been confounded in the literature. I will mainly follow the definitions given by Tang-Martinez (2001). Kin discrimination refers to the differences in the behavioural responses that an individual shows toward its kin compared to non-kin, based on conspecific labels or cues that are correlated with kinship (Tang-Martinez 2001). Kin discrimination therefore only implies to behavioural responses and not necessarily to cognitive processes. Kin recognition, on the other hand, refers to the cognitive mechanisms (i.e., neural processing) which allow animals to classify conspecifics as either kin or non-kin (see Byers & Bekoff 1986, Waldman et al. 1988).

The behavioural discrimination of kin may allow one to infer kin recognition, but the latter can never be observed directly because it is a neural process. For simplification kin discrimination and kin recognition will be used interchangeable in this thesis as in other studies (Lacy & Sherman 1983, Sherman et al. 1997), because I will examine discrimination in behaviour but not recognition abilities. A lack of discrimination in one context does not imply its absence in another (Waldman et al. 1988, Barnard & Aldhous 1991, Sherman et al. 1997) suggesting that kin discrimination operates in a context-dependent fashion.

Evidence of kin discrimination

Whether or not individuals can recognise their kin has been studied across the entire animal kingdom from social amoeba (Strassmann et al. 2000) to humans (e.g. Porter & Moore 1981). Most detailed evidence of kin recognition comes from social insects (review e.g. in Breed & Bennett 1987, Bourke & Ratnieks 1999) which are of particular interest due to their variable degree of relatedness caused by haplo-diploidy. However, broad evidence of kin discrimination has also been reported for a variety of vertebrate species, e.g., rainbowfish, Melanotaenia eachamensis, (Arnold 2000), cascades frog tadpoles, Rana cascadae (Blaustein & O’Hara 1981, 1982), long-tailed tits, Aegithalos caudatus (Hatchwell et al. 2001), house mice, Mus musculus (Kareem & Barnard 1982), white-footed mice, Peromyscus leucopus (Grau 1982), spiny mice, Acomys cahirinus (Porter et al. 1983), Belding’s ground squirrels, Spermophilus beldingi (Holmes & Sherman 1982, 1983), beavers, Castor canadensis (Sun & Müller-Schwarze 1997), and chimpanzees, Pan troglodytes (Parr & de Waal 1999, see also review for vertebrates in Blaustein et al. 1987a).

As most studies on kin recognition have focused on the discrimination of maternal (i.e., familiar) kin versus non-kin, less is known whether paternal (i.e., unfamiliar) kin can recognise each other. Only a few studies have been able to test paternal half-siblings against non-kin, e.g., house mice (Kareem & Barnard 1982), Belding’s ground squirrels (Holmes 1986b), peacocks, Pavo cristatus (Petrie et al. 1999), golden hamsters, Mesocricetus auratus (Todrank et al. 1998), and savanna baboons, Papio cynocephalus (Alberts 1999, Smith 2000). Paternal kin discrimination is most likely to evolve when paternal sibships occur in a social species. This happens (i) when unrelated females live together in a social group and (ii) when these females conceive their offspring by the same male (male reproductive skew). Both conditions are likely to be found in many primate societies.

The importance of maternal kinship in primates has been extensively studied (reviewed in more detail in chapter 5), studies including paternal kin are, however, still very limited. One captive study has investigated paternal kin discrimination in primates (Wu et al. 1980). They concluded that pigtailed macaques, Macaca nemestrina, exhibit kin recognition in the absence of familiarity, based upon their finding that unfamiliar juvenile peers prefer to sit closer to their paternal half-siblings than to non-kin. However, all subsequent studies of the same species not only failed to replicate the original findings, but inferred that familiarity rather than paternal kinship affected social preferences (Fredrickson & Sackett 1984, Sackett & Fredrickson 1987). Furthermore, captive studies of long-tailed macaques, M. fascicularis (Welker et al. 1987), and savanna baboons, Papio cynocephalus (Erhart et al. 1997) have reported that primates do not recognise paternal kin and that familiarity regulates social relationships. Nepotism (preferential treatment of kin) is thought to develop among primates as a consequence of familiarity and matrilineal, but not patrilineal, relatedness (Gouzoules & Gouzoules 1987, Walters 1987, Bernstein 1991, Chapais et al. 2001). Paternal kin discrimination has therefore been thought to be unlikely among primates for the last 20 years, but recent studies in wild baboons have resurrected the interest in the prospect that paternal kin recognition occurs in primates (Alberts 1999, Smith 2000).

 Hypothesis

In species with multiple mating, sirehood has been found to be restricted to a limited number of males each year (e.g., M. fascicularis: de Ruiter et al. 1992, P. cynocephalus: Altmann, et al. 1996, M. sinica: Keane et al. 1997, M. mulatta: Bercovitch et al. 2000). When male reproductive success is strongly skewed, peers are likely to be paternal half-siblings (J. Altmann 1979). On the other hand, peers are rarely maternal half-siblings because most female primates give birth to only a single offspring. Therefore, maternal half-siblings are at least one, but often two or more, years apart. Under this type of demographic structure, maternal half-siblings would be members of different age cohorts (i.e. non-peers), while paternal half-siblings would often be members of the same age cohort (i.e. peers).

One way of trying to distinguish familiarity from phenotype matching in primates is to examine patterns of social relationships among paternal half-siblings. If male reproduction is skewed within a season, if age mates are more familiar with each other than non-age mates, and if familiarity per se promotes social relationships, then one would expect individuals to associate more frequently with (often paternally related) peers than with non-peers. However, if relatedness per se regulates social relationships, then one would expect individuals to associate as often with maternal as with paternal kin.

 Aim of this chapter

This chapter will examine the influence of both maternal and paternal kinship, as well as age proximity, on patterns of affiliation and aggression among semi free-ranging adult female rhesus macaques. In rhesus macaques, male reproductive success is skewed (see chapter 3) and age mates play more often with each other than animals of different ages do (Janus 1992), so peers are probably frequently playing with paternal siblings. Adult female rhesus macaques from the same age cohort, therefore, have a strong likelihood of having the same sire, but the extent to which shared paternity influences adult female social relationships is unknown.

Both age proximity and paternal kinship influence mating behaviour in savanna baboons. Alberts (1999) found that paternal half-siblings were engaged in lower levels of affiliative and sexual behaviour than non-kin. Alberts also found that individuals from the same cohort tended to avoid mating with each other, but a limited sample size precluded simultaneous examination of age proximity and kinship. Here, Alberts’s hypothesis will be significantly expanded by scrutinising the influence of maternal and paternal kinship, as well as age proximity, on affiliative and aggressive relationships among adult female rhesus macaques.
Results The effect of kinship and age proximity on affiliation and aggression

Kin selection theory (Hamilton, 1964) suggests that the cost-benefit ratio of behaviours, whether they are affiliative or aggressive, will be proportional to the degree of relatedness of the actor and recipient, respectively. In other words, affiliation is expected to increase as the degree of relatedness increases, but likewise aggression is expected to decrease along with relatedness. Under conditions promoted by kin selection (Hamilton 1964) affiliation should be more pronounced among kin (both maternal and paternal) than among non-kin, whereas aggression should be less pronounced among kin than among non-kin. Here, data on affiliation and aggression are listed for maternal half-siblings, paternal half-siblings, and non-kin, the two latter being either peers or non-peers (Table 4.1). Note that maternal half-siblings can only be non-peers (see Methods). Definition on affiliative and aggressive interactions can be found in the Appendix 2. Definitions on the kin and age categories used are described in detail in the Methods.

Cell values are mean frequencies per hour (± SD) for proximity, grooming, approach (affiliative interactions), physical and non-physical aggression (agonistic interactions). Abbreviations are as follows: number of individuals tested (N), maternal half-sisters, non-peer (MS-NP), paternal half-sisters, peer (PS-P) or non-peer (PS-NP) and non-kin, peer (NK-P) or non-peer (NK-NP). Dyads that fit in more than one category (e.g., being paternal half-siblings and maternal aunt-niece) were excluded in all analyses.

Mean freq/hr

N

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

MS-NP

34

2.117 ± 1.087

0.372 ± 0.301

1.581 ± 0.754

0.017 ± 0.022

0.050 ± 0.041

PS-P

15

0.588 ± 0.419

0.088 ± 0.119

0.521 ± 0.330

0.007 ± 0.018

0.029 ± 0.072

PS-NP

19

0.450 ± 0.302

0.012 ± 0.026

0.308 ± 0.160

0.002 ± 0.007

0.024 ± 0.040

NK-P

34

0.360 ± 0.160

0.024 ± 0.030

0.327 ± 0.137

0.005 ± 0.008

0.019 ± 0.020

NK-NP

34

0.350 ± 0.111

0.011 ± 0.007

0.232 ± 0.054

0.003 ± 0.002

0.013 ± 0.006

Table 4.4.1: The effect of kinship and age proximity on affiliation and aggression

Regarding kinship, the three kin groups differed from each other in rates of affiliative and aggressive interactions. Maternal half-siblings were generally more affiliative, but also more aggressive than either paternal half-siblings or non-kin. Paternal half-siblings exhibited about 1.5 to 4 times higher rates of affiliation than non-kin, but likewise rates of aggression exceeded the one found for non-kin. Regarding age, rates of both affiliation and aggression were higher among peers than among non-peers both among paternal half-siblings and non-kin. Variation in the data was higher among paternal half-sisters than non-kin and was even more pronounced among peers, indicating that some females frequently exhibit affiliation and/or aggression towards paternal peers while others do not. For statistical tests see below.

Testing age proximity controlling for kinship

Given that peers are more familiar than non-peers (Janus 1992), one might expect that affiliative interactions are more frequent among peers than among non-peers. In addition, peers are likely to compete for the same resources at the same time of their lives, resulting in more frequent aggressive interactions among peers than among non-peers (Janus 1991). It was already suggested by Alexander (1974) that the closest competitor will also be the closest in co-operator. Therefore, affiliation and aggression between peers and non-peers were first compared in order to test for effects of age proximity on affiliation. Possible kinship effects were eliminated by restricting this comparison to non-kin. The results are summarised in Table 4.2.

Paired t-tests using data from Table 4.1. P’ represents the Dunn-Šidák correction for multiple testing undertaken for 10 simultaneous tests, i.e., per single behaviour for Table 4.2-4.4 this chapter and Table 5.3-5.4 chapter 5 (see Methods). Recall from the Methods that only P-values less than or equal to the corrected P-value (P’) indicate a significant test result which will be marked in bold.

Paired t-tests

N

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

NK-P

34

t=0.350

t=2.560

t=4.266

t=1.627

t=1.874

vs. NK-NP

P=0.728

P=0.015

P<0.001

P=0.113

P=0.070

P’=0.006

P’=0.006

P’=0.007

P’=0.009
Table 4.4.2: Testing age proximity controlling for kinship

Unrelated adult females affiliated more with their peers than with their non-peers. Peers groomed and approached each other significantly more often than non-peers, but no such difference emerged from patterns of proximity. Regarding aggression, peers tended to exhibit more non-physical aggression than non-peers, but no such difference was found for physical aggression. Overall, the results show that age proximity had a significant effect on rates of affiliation and aggression between adult female rhesus macaques. Therefore, all subsequent analyses will differentiate between peers and non-peers.

Testing maternal vs. paternal half-siblings and non-kin controlling for age proximity

Most studies on kinship in primates have compared maternal kin against maternal unrelated individuals, ignoring the proportion of paternal kin. As only few studies were able to separate paternal half-sibling from non-kin, a comparison was made between maternal and paternal half-siblings on the one hand, and maternal half-siblings and non-kin on the other hand (Table 4.3). Analyses were restricted to non-peers as maternal half-siblings are rarely peers (see Methods).

Paired t-tests using data from Table 4.1. See Table 4.2 for more details.

Paired t-tests

N

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

MS-NP

34

t=9.799

t=6.977

t=10.493

t=5.437

t=3.754

vs. NK-NP

P<0.001

P<0.001

P<0.001

P<0.001

P<0.001

P’=0.005

P’=0.005

P’=0.005

P’=0.005

P’=0.005

MS-NP

19

t=6.344

t=5.291

t=6.565

t=1.960

t=3.078

vs. PS-NP

P<0.001

P<0.001

P<0.001

P’=0.005

P’=0.005

P’=0.005

P’=0.006

P’=0.005

Table 4.4.3: Testing maternal vs. paternal half-siblings and non-kin controlling for age proximity

When compared among non-peers, maternal half-siblings were more affiliative and more aggressive towards each other in all interactions examined than either paternal half-siblings or non-kin, with the exception that maternal half-siblings showed only a trend towards more aggression than paternal half-siblings.

Testing paternal half-siblings vs. non-kin controlling for age proximity

As age proximity influenced affiliation and aggression among non-kin, it is expected that both age proximity and paternal kinship influence female social relationships. Here, a comparison was made between affiliation and aggression of paternal half-siblings vs. non-kin being either peers or non-peers (Table 4.4).

Paired t-tests using data from Table 4.1. See Table 4.2 for more details.

Paired t-tests

N

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

PS-P

15

t=-2.963

t=-2.136

t=-3.170

t=0.139

t=0.553

vs. NK-P

P=0.010

P=0.051

P=0.007

P=0.891

P=0.589

P’=0.009

P’=0.007

P’=0.010

P’=0.025

P’=0.025

PS-NP

19

t=-1.263

t=0.011

t=-2.096

t=-0.582

t=-1.244

vs. NK-NP

P=0.222

P=0.991

P=0.051

P=0.568

P=0.230

P’=0.013

P’=0.050

P’=0.013

P’=0.006

P’=0.013
Table 4.4.4: Testing paternal half-siblings vs. non-kin controlling for age proximity

To examine the effects of both paternal kinship and age proximity, affiliative and aggressive interactions were first analysed within a given birth cohorts. Among peers, paternal half-siblings were significantly more affiliative (with grooming as a trend after the Dunn-Šidákcorrection), but showed similar rates of aggression than non-kin. As a second step, affiliative and aggressive interactions were compared across birth cohorts. Among non-peers, paternal half-siblings showed a trend of approaching each other more often than non-kin, according the Dunn-Šidákcorrection, but the results were inconsistent, which was probably a consequence of including multiple ages in the non-peer category (see below). In sum, the results indicate that adult female rhesus macaques discriminate paternal half-siblings from non-kin, with paternal kin discrimination pronounced among peers in terms of affiliative interactions, but less visible regarding aggressive interactions.

Testing kinship and age proximity simultaneously

A two-way analysis of variance was applied to simultaneously examine the influence of both paternal kinship and age proximity on all affiliative and agonistic behaviours. In other words, it was investigated whether both parameters produce an interaction effect for any of the behaviours considered. Here, indices of the same focal animal with social partners of different levels of relatedness were treated as independent. (Table 4.5).

Two-way analysis of variance treating values of proximity, grooming, approach, physical and non-physical aggression of the same focal animal with social partners from different classification levels as independent. Dyads present either paternal half-siblings (PS) or non-kin (NK) with respect to kinship and either peers (P) or non-peers (NP) with respect to age. Mean, SD and sample size of the data used in this analyses are shown in Table 4.1.

ANOVA Two-way

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

Kin (d.f.=1)

PS vs. NK

F=11.071
P=0.001

F=9.395
P=0.003

F=14.855
P<0.001

F=0.115
P=0.735

F=2.352
P=0.128

Age (d.f.=1)

P vs. NP

F=2.265
P=0.136

F=17.864
P<0.001

F=19.286
P<0.001

F=3.299
P=0.072

F=0.525
P=0.470

Kin*Age (d.f.=1,98)

F=1.700
P=0.195

F=8.909
P=0.004

F=2.818
P=0.096

F=0.415
P=0.521

F=0.012
P=0.914
Table4.4.5: Testing kinship and age proximity simultaneously

When tested by means of two-way analysis of variance, both paternal kinship and peerage had a significant effect on affiliation with the exception of age on patterns of proximity. The interaction between paternal kinship and age was only found to be statistically significant for grooming. However, neither paternal kinship nor peerage had an effect on both type of aggression analysed. Referring to Table 4.1 again, a comparison of the means between paternal half-sisters and non-kin, being either peers and non-peers suggests that paternal half-sibling born into the same cohort showed consistently the highest means on all affiliative interactions, while non-kin born into different cohorts showed consistently the lowest means on all affiliative interactions. The same trend emerged for physical and non-physical aggression.

The effect of kinship and the exact age differences on affiliation and aggression

The analysis of age proximity and kinship was refined by considering the exact age difference (in years) between maternal half sisters, paternal half sisters and non-kin, using individuals either zero (i.e., peers), one, two or three years apart in age. While all 34 focal females had non-kin of all three age classes, the number of (at least one) maternal and/or paternal half siblings per focal female varied, resulting in a total N for maternal half-siblings of 43 (Table 4.6), paternal half-siblings of 39 (Table 4.7) and non-kin of 102 (Table 4.8).

Pearson correlation coefficient for focal females who had at least one maternal half-siblings (MS) with an exact age difference of either 1, 2 or 3 years apart. The exact age difference was correlated with proximity, grooming, approach, physical or non-physical aggression, respectively. Cell values are mean frequencies per hour (± SD). For 43 focal females the Pearson correlation coefficient (r p ) was calculated, treating the same focal animal with social partners from different classification levels as independent.

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

MS-1 (N=14)

1.662 ± 0.648

0.376 ± 0.401

1.173 ± 0.651

0.017 ± 0.024

0.043 ± 0.086

MS-2 (N=12)

1.966 ± 1.182

0.280 ± 0.206

1.363 ± 1.128

0.013 ± 0.023

0.056 ± 0.060

MS-3 (N=17)

2.403 ± 1.371

0.452 ± 0.417

2.002 ± 0.962

0.026 ± 0.035

0.067 ± 0.070

Pearson corr.

rp =0.278

r p =0.098

r p =0.368

r p =0.154

r p =0.141

N total =43

P=0.071

P=0.530

P=0.015

P=0.323

P=0.368
Table 4.4.6: The effect of maternal kinship and exact age differences on affiliation and aggression

As the age differences between maternal half siblings increased, rates of approach also increased, but no such association was found for aggression or grooming. A trend to increased proximity with increased age difference was also apparent.

Pearson correlation coefficient for paternal half-siblings (PS) between the exact age difference (0, 1, or 2 years apart) and proximity, grooming, approach, physical and non-physical aggression, respectively. For 39 focal females the Pearson correlation coefficient (r p ) was calculated, treating the same focal animal with social partners from different classification levels as independent. For more details see Table 4.6.

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

PS-P (N=15)

0.588 ± 0.419

0.088 ± 0.119

0.521 ± 0.330

0.007 ± 0.018

0.029 ± 0.072

PS-1 (N=15)

0.491 ± 0.320

0.026 ± 0.093

0.390 ± 0.240

0.006 ± 0.016

0.036 ± 0.049

PS-2 (N=9)

0.404 ± 0.250

0.014 ± 0.026

0.255 ± 0.149

0.000 ± 0.000

0.009 ± 0.018

Pearson corr.

r p =-0.208

r p =-0.313

r p =-0.372

r p =-0.166

r p =-0.120

N total =39

P=0.204

P=0.052

P=0.019

P=0.313

P=0.468
Table 4.4.7: The effect of paternal kinship and exact age differences on affiliation and aggression

As the exact age differences between paternal half siblings increased, rates of grooming (trend) and approaches declined consistently. However, no such association was found for proximity, physical and non-physical aggression.

Pearson correlation coefficient for unrelated females (NK) between the exact age difference (0, 1, or 2 years apart) and proximity, grooming, approach, physical and non-physical aggression, respectively. For 102 focal females the Pearson correlation coefficient (r p ) was calculated, treating the same focal animal with social partners from different classification levels as independent. For more details see Table 4.6.

Proximity

Grooming

Approach

Physical Aggression

Non-physical Aggression

NK-P (N=34)

0.360 ± 0.160

0.024 ± 0.030

0.327 ± 0.137

0.005 ± 0.008

0.019 ± 0.020

NK-1 (N=34)

0.383 ± 0.118

0.020 ± 0.020

0.313 ± 0.097

0.003 ± 0.004

0.019 ± 0.017

NK-2 (N=34)

0.390 ± 0.154

0.015 ± 0.019

0.286 ± 0.088

0.003 ± 0.005

0.019 ± 0.010

Pearson corr.

r p =0.085

r p =-0.151

r p =-0.157

r p =-0.123

r p =0.023

N total=102

P=0.394

P=0.131

P=0.114

P=0.219

P=0.820
Table 4.4.8: The effect of non-kinship and exact age differences on affiliation and aggression

The exact age differences between unrelated females had no significant effect on either affiliative or aggressive interactions.

In sum, only levels of affiliation between paternal half sisters increased gradually with decreasing age difference, while maternal half sisters tended to display the opposite pattern. Affiliation and aggression among non-kin showed no association regarding the exact age differences.

The effect of kinship and age proximity on affiliation and aggression controlling for spatial proximity

Individuals who spend a large amount of time in close spatial proximity are much more likely to interact with each other in both co-operative and competitive ways than individuals who are rarely found in spatial proximity (cf. Alexander 1974). In addition, Chapais (2001) discussed a confounding effect between spatial proximity and matrilineal kinship on the evolution of nepotism, as the persistence of mother-daughter preferential bonds into adulthood generates a structure of spatial proximity among maternal kin, which reflects genetic relatedness. Correlation between degree of relatedness and time spent in proximity was reported for Japanese macaques by Kurland (1977) and rhesus macaques by Kapsalis & Berman (1996a). As shown above, spatial proximity rates in this study also differed depending upon genetic relatedness and/or age proximity. In order to control for this possible confounding effect between proximity and relatedness, data were corrected for spatial proximity by dividing the frequency of a certain behavioural interaction (e.g., grooming) within each individual dyad by the frequency of proximity within the same dyad. The data presented above will be re-analysed after controlling them for spatial proximity (see Table 4.9). The test results which follow will be compared in respect to analyses with or without the control for spatial proximity.

Mean frequency

N

Grooming cp

Approach cp

Physical Aggression cp

Non-physical Aggression cp

MS-NP

34

0.046 ± 0.034

0.199 ± 0.057

0.002 ± 0.002

0.008 ± 0.008

PS-P

15

0.027 ± 0.033

0.250 ± 0.168

0.001 ± 0.003

0.012 ± 0.028

PS-NP

19

0.005 ± 0.011

0.226 ± 0.178

0.002 ± 0.005

0.016 ± 0.034

NP-P

34

0.012 ± 0.016

0.258 ± 0.111

0.005 ± 0.010

0.023 ± 0.045

NK-NP

34

0.005 ± 0.003

0.207 ± 0.048

0.003 ± 0.004

0.013 ± 0.007

Table 4.4.9: The effect of kinship and age proximity on affiliation and aggression controlling for spatial proximity

Cell values are mean frequencies (± SD) for grooming, approach (affiliative interactions), physical and non-physical aggression (agonistic interactions) which are controlled for proximity indicated by cp . Abbreviations as in Table 4.1.

When controlled for spatial proximity, the three kin groups differed less from each other in respect to affiliation and aggression. Maternal half-siblings still exhibited the highest rates of grooming, followed by paternal half-siblings and then non-kin, with grooming more frequent among peers than among non-peers. However, controlling for spatial proximity seemed to decrease the bias in behaviour towards maternal half-siblings.

Testing age proximity controlling for kinship and spatial proximity

Affiliation and aggression between peers and non-peers will first be compared, and possible kinship effects were eliminated by restricting this comparison to non-kin. The results are summarised in Table 4.10.

Paired t-tests

N

Grooming cp

Approach cp

Physical Aggression cp

Non-physical Aggression cp

NK-P

34

t=2.595

t=2.645

t=1.172

t=1.336

vs. NK-NP

P=0.014 (=)

P=0.012 (=)

P=0.250 (=)

P=0.191 (=)

P’=0.010

P’=0.013

P’=0.013

Table 4.4.10: Testing age proximity controlling for kinship and spatial proximity

Paired t-tests using data from Table 4.9. P’ represents the Dunn-Šidák correction for multiple testing undertaken for 5 simultaneous tests, i.e., per single behaviour for Table 4.10-4.12 of this chapter (see Methods). Recall from the Methods that only P-values less than or equal to the corrected P-value (P’) indicate a significant test result which will be marked in bold. The equal sign ( = ) indicates that the test did not change whether or not one controlled for spatial proximity, i.e., a test either remained significant or non-significant.

When controlled for spatial proximity, the results revealed that unrelated females groomed and tended to approach their peers more often than their non-peers as they did in the previous analyses, but no differences emerged with respect to aggression.

Testing maternal vs. paternal half-siblings and non-kin controlling for age proximity and spatial proximity

Analyses were restricted to non-peers, comparing maternal half-siblings with non-kin on the one hand and maternal half-siblings with paternal half-siblings on the other hand (Table 4.11).

Paired t-tests using data from Table 4.9. See Table 4.10 for more details. Note, no equal sign in a non-significant result indicates that this test was significant before.

Paired t-tests

N

Grooming cp

Approach cp

Physical Aggression cp

Non-physical Aggression cp

MS-NP

34

t=7.113

t=-0.843

t=-0.897

t=-3.131

vs. NK-NP

P<0.001 (=)

P=0.406

P=0.376

P=0.004 (=)

P’=0.010

P’=0.017

P’=0.017

P’=0.010

MS-NP

19

t=4.667

t=-0.545

t=0.420

t=-1.051

vs. PS-NP

P<0.001 (=)

P=0.593

P=0.680 (=)

P=0.307

P’=0.013

P’=0.050

P’=0.050

P’=0.025
Table 4.4.11: Testing maternal vs. paternal half-siblings and non-kin controlling for age proximity and spatial proximity

When compared among non-peers and controlled for spatial proximity, maternal half-siblings still differed significantly in grooming and non-physical aggression from non-kin as in the previous analyses, but the previous differences found in approach and physical aggression did not hold when data were controlled for spatial proximity. On the other hand, maternal half-siblings still differed significantly in grooming from paternal half-siblings than in the previous analyses, but the previous differences found in approach and non-physical aggression did not hold when data were controlled for spatial proximity.

Testing paternal half-siblings vs. non-kin controlling for age proximity

Here, the comparison is re-analysing affiliation and aggression among paternal half-siblings vs. non-kin being either peers or non-peers (Table 4.12).

Paired t-tests using data from Table 4.9. See Table 4.10 for more details. Note, no equal sign in a non-significant result indicates that this test was significant before.

Paired t-tests

N

Grooming cp

Approach cp

Physical Aggression cp

Non-physical Aggression cp

PS-P

15

t=1.192

t=-0.900

t=-1.655

t=-1.148

vs. NK-P

P=0.253 (=)

P=0.383

P=0.120 (=)

P=0.270 (=)

P’=0.025

P’=0.013

P’=0.010

P’=0.017

PS-NP

19

t=-0.053

t=0.774

t=-0.575

t=0.675

vs. NK-NP

P=0.958 (=)

P=0.449 (=)

P=0.572 (=)

P=0.508 (=)

P’=0.050

P’=0.025

P’=0.025

P’=0.050
Table 4.4.12: Testing paternal half-siblings vs. non-kin controlling for age proximity and spatial proximity

Recall from the previous analyses, that the comparison among peers revealed that paternal half-siblings tended to groom and approach each other more often than non-kin. When controlled for spatial proximity, paternal half-siblings showed similar rates of affiliation and aggression as non-kin, both among peers and among non-peers. In other words, relative to proximity, paternal half-siblings and non-kin interact with each other at similar rates.

Testing kinship and age proximity simultaneously

Referring to Table 4.9 again, a comparison was made including paternal kinship and age proximity simultaneously in order to investigate whether both parameters interact with each other (Table 4.13).

Two-way analysis of variance treating values of grooming, approach, physical and non-physical aggression of the same focal animal with social partners from different classification levels as independent. Values were controlled for spatial proximity. Dyads present either paternal half-siblings (PS) or non-kin (NK) with respect to kinship and either peers (P) or non-peers (NP) with respect to age. Mean, SD and sample size of the data used in this analyses are shown in Table 4.9.

ANOVA Two-way

Grooming cp

Approach cp

Physical Aggression cp

Non-physical Aggression cp

Kin (d.f.=1)

PS vs. NK

F=4.183
P=0.044 (=)

F=0.040

P=0.841

F=2.840

P=0.095 (=)

F=0.348

P=0.556 (=)

Age (d.f.=1)

P vs. NP

F=17.858
P<0.001 (=)

F=2.118

P=0.149

F=0.432

P=0.513 (=)

F=0.173

P=0.678 (=)

Kin*Age (d.f.=1,98)

F=4.686
P=0.033 (=)

F=0.276

P=0.600 (=)

F=0.795

P=0.375 (=)

F=1.161

P=0.284 (=)
Table 4.4.13: Testing kinship and age proximity simultaneously

Controlling for spatial proximity had no effect on grooming which stayed significant when tested for paternal kinship, age and their interaction, but influenced the outcome of other interactions measured. In other words, regardless of whether proximity was controlled for or not, kinship and age proximity interact to regulate grooming frequency between females.

In summary, when controlled for spatial proximity peers still affiliated with each other more than with non-peers. Maternal half-siblings still groomed each other more than either non-kin or paternal half-siblings, and exhibited more non-physical aggression than non-kin.

The correlation between affiliation and aggression

Finally, an attempt was made to correlate affiliation with aggression for the three kin categories investigated in this chapter. This comparison was performed to test Alexander’s (1974) suggestion that pairs who are most affiliative are also most aggressive towards each other. The three kin groups will be analysed among non-peers (Table 4.14) and two kin groups among peers (Table 4.15).

Spearman’s rank correlation coefficient between affiliation (proximity, grooming or approach) and aggression (physical or non-physical) among non-peers including maternal half-siblings (MS-NP), paternal half-siblings (PS-NP) or non-kin (NK-NP).

MS-NP (N=34)

PS-NP (N=19)

NK-NP (N=34)

Physical Aggression

Non-physical Aggression

Physical Aggression

Non-physical Aggression

Physical Aggression

Non-physical Aggression

Proximity

rs=0.018

P=0.920

rs=-0.129

P=0.467

rs=0.033

P=0.894

rs=0.269

P=0.266

rs=0.454
P=0.007

rs=0.312

P=0.073

Grooming

rs=-0.015

P=0.931

rs=0.125

P=0.482

rs=0.287

P=0.234

rs=-0.083

P=0.734

rs=0.289

P=0.097

rs=0.444

P=0.008

Approach

rs=-0.084

P=0.636

rs=-0.052

P=0.768

rs=0.339

P=0.156

rs=0.149

P=0.542

rs=0.267

P=0.126

rs=0.298

P=0.087
Table 4.4.14: The correlation between affiliation and aggression among non-peers

Among non-peers, a correlation between affiliation and aggression was not apparent, neither for maternal half-siblings nor paternal half-siblings, but was evident for non-kin. Females directed significantly more physical aggression to non-kin, the more these females shared spatial proximity with each other. Females also directed more non-physical aggression to non-kin, the more these females groomed each other. These results revealed a positive relation between affiliation and aggression only among non-kin, not among maternal and paternal kin.

Spearman’s rank correlation coefficient between affiliation (proximity, grooming or approach) and aggression (physical or non-physical) among peers being either paternal half-siblings (PS-P) or non-kin (NK-P).

PS-P (N=15)

NK-P (N=34)

Physical Aggression

Non-physical Aggression

Physical Aggression

Non-physical Aggression

Proximity

rs=0.590

P=0.021

rs=0.079

P=0.779

rs=0.301

P=0.083

rs=-0.041

P=0.819

Grooming

rs=0.431

P=0.109

rs=-0.096

P=0.735

rs=0.291

P=0.094

rs=0.017

P=0.925

Approach

rs=0.592
P=0.020

rs=0.064

P=0.821

rs=0.309

P=0.075

rs=0.053

P=0.766
Table 4.4.15: The correlation between affiliation and aggression among peers

Among peers, paternal half-siblings directed significantly more physical aggression towards each other, the more they shared spatial proximity or approached each other. Maternal half-siblings could not be analysed because they are not peers (see Methods). Unrelated peers showed no association between affiliation and aggression.

In summary, the results revealed no correlation between affiliation and aggression among maternal half-siblings. Although maternal half-siblings exhibited the highest frequency of both affiliation and aggression, these measurements did not correlate with each other. In other words, some focal females showed high levels of affiliation, but not of aggression, towards their maternal half-siblings, while other focal females showed high levels of aggression, but not of affiliation, towards their maternal half-siblings. The data support Alexander’s hypothesis (1974) that high affiliative individuals are also engaged in a high level of aggression, but only among some kinship and age categories.

Discussion

The results of this chapter confirm that the closest social bonds among adult female rhesus macaques follow maternal relatedness. Maternal half-siblings are both more affiliative and aggressive towards each other than either paternal half-siblings or non-kin. The present study also provides the first evidence that both paternal relatedness and age proximity exert a significant impact upon affiliation patterns among adult female rhesus macaques. Paternal half-siblings prefer to affiliate with each other in comparison with non-kin, but patterns of aggressive activity show no kin discrimination according to paternal relatedness. Among paternal half-siblings, rates of affiliation, but not of aggression, increased with decreasing age difference whereas among unrelated females neither patterns of affiliation nor aggression were influenced by exact age difference. Among maternal half-siblings, increasing age difference had no impact on patterns of aggressive interactions, but decreasing exact age differences were associated with significant decreases in rates of approach. Data of this study demonstrate that (i) adult female rhesus macaques are capable of discriminating between paternal kin and non-kin, with kin discrimination more pronounced among peers than among non-peers and inversely related to age proximity, (ii) paternal kin discrimination is context-dependent because it is exhibited during affiliative, not aggressive, interactions, (iii) kin discrimination is asymmetric because maternal and paternal half-siblings interact differently, and (iv) both familiarity and phenotype matching probably drive kin discrimination in rhesus macaques. These points will now be discussed in more detail.

Peer effect

The bias towards peers in both affiliation and aggression (hereafter: peer effect) was already described among immature rhesus macaques for both sexes (Janus 1991, 1992). Peers were more affiliate and more aggressive towards each other than non-peers. Strong affiliative partners also exhibit more aggression, but aggression directed towards strong affiliate partners was less severe (ibid.). This suggests that familiarity not only favours affiliation, as aggression among age mates may reflect that they compete for the same resources at the same time in their life. In another study on the same population as studied here, Colvin & Tissier (1985) distinguished between strong and weak relationship of 3-year-old male rhesus macaques with strong peer relations defined as those in which each partner was involved for more than 15% of the total peer proximity and with reciprocity among strong but not among weak peer relations. However, paternity was not available in this earlier study, but paternal kinship was also not suggested as a likely explanation for the differences found between strong and weakpeer relations.

Evidence for a peer effect has also been found in other species. Play among Japanese macaques was most frequently observed between peers at the age of one (Koyama 1985). In general, play was observed more often than expected among age mates until the age of 4 years, but also among individuals with disparity of age less than one year (Koyama 1985). Familiarity among play mates could influence mating decisions as adults. Alberts (1999) found among savanna baboons that peers were less likely to be engaged in sexual consort than non-peers, suggesting that familiarity could result in mating avoidance. Additional support for my findings comes from Smith (2000), studying the same baboon population as Alberts. Smith (2000) found that unrelated peers are more affiliative than unrelated non-peers. These studies support the hypothesis that familiarity among individuals can arise through association in early development by at least two alternatives: (i) mothers mediating familiarity among her offspring (which are maternal half-siblings) and (ii) age proximity is mediating familiarity among age mates (including both paternal related and unrelated peers) which spend more time together than individuals of different ages do due to the formation of play groups among peers.

It should be emphasised that familiarity among peers cannot explain why paternally related peers are more affiliative than unrelated peers. In other words, as both groups of individuals are likely to be familiar to a similar extent, the higher frequency of affiliation among paternal half-sisters than among unrelated females must have been evolved through a different mechanism than familiarity. Controlling for the age differences this test was indeed restricted to compare different degrees of genetic relatedness. Smith (2000) noted that being similarly aged has two important consequences in promoting familiarity. First, barring death and group fission, female peers are likely to interact their entire lives (especially in species where females do not leave their natal group), while very different aged individuals do not. Second, female peers go through important life history stages at similar times, e.g., infancy, menarche, pregnancy, motherhood, while differently-aged females do not. This applies to all age cohort members, regardless of the variation in relatedness.

The behavioural differences found between paternal sisters and non-kin of different ages (non-peers) were not very clear comparing single behaviours, as the only significant difference found was that paternal half-siblings approached each other more other than non-kin. In contrast, combining affiliative measurements to a single affiliation index (cf. Widdig et al. 2001) there was a significant differences found between peers and non-peers. The advantage of using single behaviours is that they can show context-dependent kin discrimination (cf. Widdig et al. 2002) whereas the advantage of using an affiliation index is that it can reveal general patterns of behaviour. The less pronounced differences found among non-peers might be due to the fact that the non-peer category include individuals of multiple ages, i.e., unrelated non-peers varied in age between 1-10 years, while paternal sisters being non-peers varied in age between 1-3 years (with the exception of one pair of paternal half-siblings being 4 years apart in age). Therefore the exact age difference was used showing that affiliation is increasing the closer paternal half-siblings are in age, but the reverse effect appeared for maternal half-siblings, affiliation decreased the closer they are in age. However, unrelated females show no association between exact age difference and affiliation.

Context-dependent kin discrimination

Another important result from the present study is that female rhesus macaques bias their behaviour towards paternal half-siblings, but they did so in only some contexts, e.g., affiliation and not aggression. Context-dependent kin discrimination (Waldman et al. 1988, Barnard & Aldhous 1991, Sherman et al. 1997) was reported among other vertebrates. For example, spadefoot toad tadpoles, Scaphiopus bombifrons, occur in two morphs with omnivores preferentially associating with siblings and carnivores preferentially associating with non-siblings (Pfennig et al. 1993). Carnivores avoid eating kin, but they become less selective when hungry, suggesting that their level of kin discrimination is context-dependent (ibid.). While female white-footed mice, Peromyscus leucopus, prefer the odour of males of intermediate relatedness when they are in oestrous, non-oestrous females show no preference (Keane 1990). In savanna baboons Alberts (1999) found that pairs of paternal brothers and sisters were just as likely to be in sexual consort as unrelated pairs, but consorts of paternal half-siblings were less afffiliative and sexual than that of non-kin. In the same baboon population Smith (2000) reported that paternal half-sisters of different age groomed each other more than unrelated females of different age, but no distinction was found with respect to the nearest neighbour while resting.

Kin discrimination seems to be detected only when behaviours most linked to fitness are examined under relevant conditions. For example, studies on prairie voles, Microtus ochrogaster, had found a breakdown of incest avoidance after 8-15 days of sibling isolation, but a re-evaluation using amicable and agonistic interactions of same sex individuals suggested that sibling recognition is still present when isolation was less than 20 days (Paz y Mino & Tang-Martinez 1999). Again, this emphasises the importance of studying multiple behaviours under different conditions. However, Paz y Mino & Tang-Martinez (1999) suggested from their findings that kin recognition can be detected at different time intervals using different approaches. The current finding that affiliation, not aggression, is associated with kin discrimination in adult female rhesus macaques would be expected if co-operative interactions had a greater impact on fitness than competitive interactions (Jolly 1999).

Asymmetry in behaviour between maternal and paternal half-siblings

The data of the present study show a strong asymmetry in both affiliation and aggression between maternal and paternal half-siblings in favour of the former even though they share on average the same degree of relatedness. Given that maternal half-sibling are of adjacent rank whereas paternal half-sibling and non-kin may considerably vary in dominance rank (unpubl. data), the bias towards maternal kin may be a by-product of the attraction to similar ranking females as suggested by de Waal (1991), Kapsalis & Berman (1996a,b) and Chapais et al. (1991, 1994). Silk also noted (2001) that paternal kinship may play a less salient role than maternal kinship because there is always some degree of uncertainty about paternity. In addition, because male reproductive skew is fare from being perfect sharing the same peer group does not necessary mean that individuals also share the same sire.

However, the only study available which also compares maternal half-siblings vs. paternal half-siblings and paternal half-siblings vs. non-kin is the one on wild baboons in Amboseli by Smith (2000) using similar affiliative measurements as my study. Like the present study female baboons also exhibit paternal kin discrimination, as paternal half-sister affiliate more with each other than non-kin. However, in contrast to the present study, baboons did not differentiate between maternal and paternal half-sisters, as they were equally likely to affiliate. Interestingly, in one measurement (total number of counts of affiliation) paternal half-siblings directed more affiliation towards each other than maternal half-siblings, i.e., the opposite trend than found in the present study. There are at least two demographic explanations for the difference between the baboon and the rhesus study. First and probably most important, Smith also found a bias towards peers, but she did not distinguish peers from non-peers. In other words, she did not control for age when testing kinship between maternal and paternal half-siblings probably due to limited sample size. This adds familiarity through peerage to genetic relatedness in the case of paternal half-siblings which maternal half-siblings, typically of different ages, are missing. Second, Chapais (2001) noted that kin-bias should strongly be affected by the number of kin per kin classes. Thus population difference may be responsible for the discrepancy, as Smith pooled data over three small group with paternal half-sisters being equally distributed among groups, but a single family from one group disproportionately contributed data for maternal half-sibling. The number of maternal half-siblings per individual was also much higher in the present study than in the baboon study. However, Smith herself pointed out some confounding factors that may influence her analyses as the relative proportion of close kin to non-kin, i.e., the female with the highest frequency of grooming found for all pairs of paternal half-sisters had no maternal kin in the group and in addition her paternal half-sister was the only peer in the group. In contrast, the present study used always paired tests, meaning that each focal female had at least one kin per kin class tested.

Mechanisms of kin recognition

One prerequisite for kin selection is that individuals have the ability to discriminate between kin and non-kin (Grafen 1990). As outlined above, kin discrimination among primates is most likely to arise by one of the two following mechanisms. First, individuals learn the phenotypes of related individuals during early development and later discriminate these familiar relatives from unfamiliar individuals, i.e., familiarity due to prior association (Walters 1987, Tang-Halpin 1991). Second, individuals learn their own phenotypes and/or those of their familiar kin, and later compare or match the phenotypes of unknown individuals with this learned template, i.e., phenotype matching (Holmes & Sherman 1983, Lacy & Sherman 1983). Unfortunately, phenotype matching is often used as a proxy for genetic mechanisms, and familiarity denotes learning processes and some studies did not separate familiarity and relatedness. For example, recent work by Tang-Martinez (2001) suggests to combine familiarity via association and phenotype matching to one mechanism as both are based on learning. However, it is still under discussion whether these two mechanisms are mutually exclusive or overlapping (Heth et al. 1998). Although both mechanisms involve a comparison between encountered phenotypes and recognition templates, familiarity leads to recognition of previously encountered familiar individuals, whereas phenotype matching permits recognition of unfamiliar kin, through generalisation of learned templates (Holmes & Sherman 1982, Sherman et al. 1997). This distinction suggested to Mateo (2002) has implications for the evolution of kin-directed behaviours because phenotype matching permits more refined kin-differentiated behaviour than familiarity. Hamilton himself (1964, p. 22) proposed that one possible mechanism mediating kin selection could be “familiarity of appearance…being (that) relatives must tend to look alike…”. Heth et al. (1998) have expanded this concept by noting that association patterns during ontogeny, i.e., familiarity, could be the critical variable fostering phenotype matching mechanisms. It seem possible that familiarity and phenotype matching can be alternative mechanisms under some conditions as suggested by Heth et al. (1998), but the data of the present study suggest that paternal kin discrimination arises from an interaction between familiarity and phenotype matching that is expressed in an asymmetric, context-dependent fashion. In other words, the primary mechanisms used for paternal kin discrimination is familiarity through peerage and that phenotype matching is used to distinguish paternal half-siblings from non-kin.

The most frequently mentioned mechanism of kin recognition across species is phenotype matching, which depends upon comparing shared traits within a lineage to a reference template using visual, vocal or olfactory cues (Holmes & Sherman 1983, Lacy & Sherman 1983, Blaustein et al. 1987a,b, Tang-Halpin 1991). Mechanisms of kin discrimination are probably not identical across species. Differences are to be expected because recognition mechanisms must function within the confines set by a species phylogenetic history and reflect key sensory modalities used in communication, as well as mating and demographic structure. Co-operation between adult female rhesus macaques seems to be promoted by paternal kinship, but the mechanisms underlying paternal kin discrimination found in this study remain unknown.

One possible explanation could be that females nurture co-operation among their patrilineally related offspring by encouraging infants to play with offspring of females who have mated with the same male. Maternal affiliation patterns partly drive the development of infant affiliation patterns among peers (Berman & Kapsalis 1999) so that, if mothers who have conceived infants from the same male were closely associated, their offspring could develop affiliations with same-age paternal relatives. The hypothesis of maternal behavioural cues was tested by comparing mothers who were unrelated non-peers, but whose infants were either same-age paternal half-siblings or same-age non-kin. In the study population, however, mothers of paternal half-siblings who were peers had proximity scores similar to mothers of non-kin peers, as no difference emerged from these two sets of mothers (mean proximity per hour (± SE) between mothers of same-age paternal half-siblings 0.464 ± 0.065 vs. mothers of non-kin peers 0.451 ± 0.027, paired t=-0.233, d.f. 15, P=0.819). Therefore, mothers do not appear to mediate affiliation between their infants as a function of shared paternity. Instead, adult females tend to associate with their own kin and/or peers regardless of the identity of the sire of their offspring.

A second possible explanation for paternal kin discrimination could be that dominance rank influences affiliation patterns. In rhesus macaques, as in other cercopithecine primates, maternal half-sisters do not only affiliate with each other at high rates, but tend to occupy adjacent dominanceranks within a troop (Chapais & Schulman 1980). In order to test whether dominance rank was a confounder in the analysis, the mean rank difference between focal females and their paternal half-siblings was compared with the mean rank difference between focal females and their non-kin, controlling for age proximity. No significant differences in mean rank difference were observed, neither for peers (mean rank difference (± SE) between paternal half-siblings=33 ± 5 vs. non-kin=32 ± 3, Wilcoxon-test, z=-0.031, d.f. 14, Pe=0.987) nor for non-peers (paternal half-siblings=35 ± 3 vs. non-kin=32 ± 1, Wilcoxon-test, z=-0.322, d.f. 18, P=0.748). However, including all females in the analysis could bias results if younger females were dependent in rank upon their mothers. In order to control for this potential bias, the analysis was repeated by excluding younger females for which the social rank could still be dependent upon their mothers’ rank (N=42). Of the nineteen focal subjects with paternal half-sibling, non-peers, three of them were removed from the analysis because all of their paternal siblings were immature animals. Results remained consistent with no significant differences noted, neither among peers (mean rank difference (± SE) between focal and paternal half-siblings=18 ± 3 vs. 17 ± 1 for non-kin, Wilcoxon-test, z=-0.114, d.f. 14, Pe=0.934) nor among non-peers (paternal half-siblings=20 ± 2 vs. non-kin=18 ± 1, Wilcoxon-test, z=-0.724, d.f. 15, P=0.469). In conclusion, although paternal half-siblings and non-kin had a similar rank in relation to the focal females, the latter still affiliated more with paternal half-sisters than non-kin, implying that differences in affiliation are unlikely to be caused by differences in relative dominance rank.

A third explanation for paternal kin discrimination could be that the differences found between paternal half-siblings and non-kin are due to differences in age proximity among peers. Recall from the Methods that individuals born into the same birth season, i.e., peers can be born on the same day or up to 6 months apart. Comparing the mean birthdate difference between peers revealed the opposite trend. Peer non-kin were actually closer in age to each other than peer paternal half-siblings (mean birthday difference in days (± SE) between peer paternal half-siblings=45 ± 7 vs. peer non-kin=33 ± 6, paired t=2.199, d.f. 14, P=0.045). Therefore, the preference of paternal half-siblings for each other within the peer group is not a consequence of close proximity of date of birth.

A fourth explanation for paternal kin discrimination could be phenotype matching, using shared characteristics within lineages, such as appearance, odour, or vocalisations, against a reference template (Holmes & Sherman 1983, Lacy & Sherman 1983, Tang-Halpin 1991, Dawkins 1982, Blaustein et al. 1987a,b). The data presented here are compatible with the phenotype matching hypothesis because paternal kin discrimination was most pronounced among peers, but was still trendy among non-peers. However, phenotype matching cannot account for the preference of peers over non-peers (peer effect), and the observed interaction between kinship and age proximity suggests that familiarity among age mates also contributes to paternal kin discrimination.

Given the crucial importance of vision among cercopithecine primates, the most likely sensory mechanism presenting cues for phenotype matching is visual recognition, as suggested for chimpanzees by Parr & de Waal (1999). However, in contrast to chimpanzees, Old World monkeys have consistently failed mirror self-recognition tests (Gallup 1997, De Veer & van den Bos 1999) and long-tailed macaques, M. fascicularis, do not seem to perceive physical resemblance between relatives (Dasser 1988). Furthermore, for the age proximity effect to be regulated by visual cues, adult females would need to be able to distinguish same-age peers from those who are a couple of years older or younger than themselves. Therefore, it seems to be unlikely that paternal kin discrimination in rhesus macaques is visually-mediated.

Among rodents, olfactory cues have been implicated in kin discrimination (e.g., Todrank et al. 1998, Mateo & Johnston 2000). New World monkeys scent mark and discriminate both sex and reproductive state of conspecifics (e.g., Converse et al. 1995, Smith & Abbott 1998), but Old World monkeys do not scent mark and have a poorly developed olfactory sense (Martin 1990). Most olfactory inspection among rhesus macaques occurs when males sniff the anogenital region of females (Bercovitch, pers. comm.), but sniffing is not necessary to perceive odour. Olfactory cues among mice are used to detect dissimilarity in the major histocompatibility complex (MHC) region as a means of mate choice (Yamazaki et al. 1976) and humans prefer the odour of MHC-dissimilar individuals (Wedekind et al. 1995, Wedekind & Füri 1997). However, unlike the situation with house mice, no reproductive advantage was found in the same population studied here as dissimilar and similar mates in terms of MHC type (Mamu-DQB locus) were equally likely to be found, even though heterozygous males overall sired more offspring than homozygous males (Sauermann et al. 2001). It seems unlikely that females adjust social behaviour with female conspecifics according to MHC type when their reproductive success is not dependent upon adjusting mating behaviour according to MHC type. However, inbreeding avoidance may still be an outcome of kin recognition that is depended upon other MHC loci and olfactory discrimination can still regulate female social relationship, even though self-recognition by odour has not yet been shown in rhesus macaques.

Auditory cues provide signals indicating maternal relatedness in both rhesus macaques (Hauser 1996, Rendall et al. 1996) and Savannah baboons (Cheney & Seyfarth 1999). However, in these species matrilineal relatedness corresponds with familiarity (Grafen 1990) and these studies did not separate maternal unrelated individuals into paternal kin and non-kin. For vocal cues to mediate kin discrimination among adult females, they would need to match paternal half-sibling utterances to self utterances, and distinguish them from non-kin calls, but auditory matching to self is difficult because the sounds emitted by an individual are perceived differently by the sender than they are by the receiver. Again, paternal kin discrimination in rhesus macaques seems therefore to be unlikely for mediating via auditory channels.

In conclusion, in the case of paternal kin discrimination discussed here, individual A is interacting more with (paternal half-sibling) B than with (non-kin) C, which must involve a self-referential phenotype matching system. No studies of non-human primates have yet examined prospects for matching-to-self in either the auditory or olfactory realms, and a “phylogenetic gap” between monkeys and apes is apparent in the visual realm because the former fail to display self-recognition (de Veer & van den Bos 1999, Hauser et al. 2001). Hence, although the most likely sensory modality mediating kin discrimination in rhesus macaques is vision, unless this species is capable of matching-to-self, then phenotype matching using visual cues cannot explain the current findings. Heth et al . (1998) proposed that self-referential phenotype matching provides the best means for guiding kin discrimination, but no solid evidence for such a process exists among non-hominoid primates.

As a result of the present study a novel mechanism of phenotype matching in primates was suggested (Widdig et al. 2001). We hypothesise that phenotype matching in rhesus macaques, and other cercopithecine species, is guided by behavioural traits, such as personality and temperament rather than by morphological or physiological attributes. Non-human primates develop and display distinct personality profiles (Clarke & Boinski 1995, Bolig et al. 1992, Sapolsky & Ray 1989). In rhesus macaques, some personality traits, such as increased impulsivity and aggressiveness, are closely associated with diminished concentrations of cerebrospinal fluid monoamine metabolites, which have a significant paternal genetic component (Higley et al . 1993, Clarke et al . 1995) and are fairly stable throughout life (Higley et al . 1996, Stevenson-Hinde et al . 1980). About 30-50% of the variance in personality traits among people is thought to be due to genetic factors (Kagan 1994, Loehlin 1992). Hence, shared paternally inherited personality attributes could be modulating social relationships and provide a mechanism fostering behavioural phenotype matching. One prediction of this hypothesis is that if individuals choose social partners on the basis of age and personality traits, and if these traits are partly determined by paternal genes, then preferred social partners will share paternity and age proximity more often than expected by chance alone. Comparing the relatedness and personalities of playmates within and across cohorts should be the next step in testing this hypothesis.

Summary

Paternal kin discrimination influences the structure of social relationships in female rhesus macaques as a function of both age proximity and shared paternity. It is proposed that paternal relatedness and age proximity regulate the development of social relationships through an ontogenetic process of phenotype matching using behavioural cues modulated by inherited personality traits. In cercopithecine primates both mating and social preferences could be a function of paternal kinship and age proximity acting via a mechanism involving personality phenotype matching. Baboons tend to avoid mating with paternal kin of the same cohort (Alberts 1999), and rhesus macaques prefer associating with paternal kin of the same age (this study). The proposed mechanism for paternal kin discrimination suggests that fitness consequences resulting from kin selection favouring nepotism (Hamilton 1967, West Eberhard 1975) and mate selection favouring inbreeding avoidance share a common foundation. In both circumstances, female reproductive success is maximised through the development of kin recognition devices that emerge during ontogeny when females develop the capacity to match their own personality template with that of conspecifics who share the same sire. Kin discrimination in primates is expressed in a context-dependent fashion with those behaviours most relevant to reproductive success the ones that extract the expression of kin discrimination.