Historically seen, this study is one of the first on free-ranging primates focusing on paternal kinship, because lacking paternity data most studies on kinship were limited to maternally related individuals. Why do we still lack studies investigating paternal kinship in primates? Firstly, technical difficulties to establish paternity from blood, hairs or even faeces samples had to be overcome (see Launhardt 1998, Gerloff et al. 1999, Smith et al. 2000). Secondly, for species with a long life span as primates it takes a considerable amount of time (after accumulate demographic data to reconstruct maternal kinship) to establish paternity for a whole group taking into account that males from outside the group are also likely to sire offspring (Berard et al. 1993, Soltis et al. 2001, this study). Thirdly, when paternally related individuals were combined with non-kin into “maternally unrelated individuals” in studies lacking paternity data, there was consistently reported a strong bias towards maternal kin in a number of species which seems to imply that paternal kinship is not likely to be important. As this study showed this might be true when comparing maternal half-siblings with paternal half-siblings, but there is still a difference when comparing paternal half-siblings with non-kin.

There was (and probably still is) an additional problem. The first evidence in favour of paternal kin recognition among primates (Wu et al. 1980) was contradicted by all subsequent studies (Fredrickson & Sackett 1984, Sackett & Fredrickson 1987, Welker et al. 1987, Erhart et al. 1997). Unfortunately, it was rarely questioned (but see Walters 1987), whether individuals isolated from their mothers soon after birth, reared in peer groups, tested with strangers rather than group members may have lacked important developmental stages, which might be necessary to recognise paternal half-siblings. Ever since most primatologist seemed to be convinced that nepotism among primates is a consequence of familiarity and therefore paternal kin cannot recognise each other (e.g., Gouzoules & Gouzoules 1987, Walters 1987, Bernstein 1991, Chapais et al. 2001), even though there is evidence from non-primate species that they can recognise paternal kin (e.g., Kareem & Barnard 1982, Holmes 1986b, Todrank et al. 1998, Petrie et al. 1999). However, kin discrimination should be expected when individuals live with their kin and non-kin together, exhibiting behaviours costly to the actor (cf. Silk 2002). Recent studies on free-living baboons and macaques have resurrected the interest in the prospect of paternal kin recognition (Alberts 1999, Smith 2000, Widdig et al. 2001, 2002), but more studies are still needed.

It is difficult to assess the influence of kinship when similarity in relatedness is associated with a high level of familiarity (measured as spatial proximity) as well as similarity in rank. These are variables which maternal kin commonly share and which are likely to be confounded. For example, it has been suggested that female primates are attracted to others most resemble in rank, age and kinship (de Waal & Luttrell 1986, de Waal 1991), but Kapsalis & Berman (1996b) could demonstrate using the matrix correlation method that maternal kinship is the primary factor influencing female social relationship, not similarity in rank.

What are likely confounding factors among paternal half-siblings? As most paternal half-sibling tend to be members of the same birth cohort or at least similar in age, age may be a confounding factor with respect to paternal kinship. However, one can separate this by controlling either for age or kinship resulting in four test groups: paternal half-siblings being either peers or non-peers and non-kin being either peers or non-peers which the present study was able to do. In addition, the mean rank difference between paternal half-siblings and non-kin did not differ, but is likely to be larger than among maternal half-siblings which are always of adjacent rank (Widdig et al. 2001, chapter 4).

Smith (2000) discussed that kin bias among paternal half-siblings should be even stronger than among maternal half-siblings due to a higher degree of relatedness, a closer similarity in age proximity and the larger difference in rank with the latter reflecting the attraction for higher-ranking females. I agree with the first two points, but not with the third point she made for the following reason. Assumed, a pair of paternal half-sisters which is very distant in rank, one high-ranking in the centre of the group, one low-ranking in the periphery of the group. First, they need to overcome the problem of spatial distance in order to interact. Following Smith (2000) we would only expect the lower-ranking paternal half-sister to approach and groom her higher-ranking paternal half-sister, not vice versa. With respect to coalition formation it seems even more unlikely that the lower-ranking paternal half-sister will intervene in favour of her higher-ranking paternal half-sister. In case that the higher-ranking paternal half-sister outranks her opponent, she does not require the support of her lower-ranking paternal half-sister, while in case the higher-ranking paternal half-sister does not outrank her opponent, her lower-ranking paternal half-sister may not intervene as she risks retaliation. In fact, the variation in affiliation among dyads of paternal half-siblings in the present study was probably due to large rank distance within some dyads (unpubl. data) an effect which can even be more pronounced with increasing group size as on Cayo Santiago.

What should future work focus on? Firstly, data from other populations and other species than the one studied are needed as variation is already seen when comparing the baboon and the rhesus studies probably reflecting differences in population.

Secondly, it would be worth investigating whether paternal kin discrimination also works among males. When a male grows up in his natal group he is as likely as a females to be paternally related with his peers. Assuming that a mechanism for paternal kin recognition develops in early development, males are expected to learn this, too, even though they leave their group around puberty to migrate elsewhere. But even then, studies on long-tailed macaques reported that males tend to migrate with their peers which are like to be their paternal half-siblings (de Ruiter & Geffen 1998). Dispersal is associated with a higher probability of mortality (increased predation risk) while losing chances to mate (Alberts & Altmann 1995). Are paternal half-brothers who leave their natal group together more successful in entering and reproducing in the new social group than unrelated males migrating together? Or if males migrate alone, do they join the group that they have paternal half-brothers in as reported for maternal half-brothers (Meikle & Vessey 1981)? Likewise, do males avoid to immigrate into groups that consist of a large proportion of paternal half-sisters as inbreeding via the paternal line is likely to be costly, too. And when reproducing successfully, do males recognise their offspring to avoid infanticide or mating with their daughters? Some studies suggested that fathers preferentially interact with their offspring (e.g., Stein 1984, Taub 1984), but studies using paternity data do not support this hypothesis (Paul et al. 1996). In species with infanticide as the Hanuman langurs, only genetic fathers or resident males at the time of the infants conception were likely to protect infants, but never males who immigrated after the infants conception suggesting that males count copulations with fertile females as clues for paternity (Borries et al. 1999). Alberts (1999) reported that pairs of paternal half-siblings exhibited less sexual behaviour than unrelated pairs. In contrast, Barbary macaques showed no mating avoidance among paternal kin (father-daughter and brother-sisters), but only 2 out of 62 potential inbred infants were found via the paternal line (Kuester et al. 1994).

Thirdly, more evidence is required for females. For adult baboons and macaques who showed evidence for paternal kin recognition it would be worth investigating when they start learning to distinguish between paternally related and unrelated individuals. In addition, which cue do they use, is it smell, vision, or maybe personality matching as hypothesised by the present study. The extent of context-dependent kin discrimination as well as the asymmetry in behaviour found between maternal half-siblings and paternal half-siblings should be compared across populations.

Fourthly, it would be worth studying paternal kinship effects related to group fission in more detail as it was reported that female baboons are more likely to end up in the same fission group with their paternal half-sibling peer than with their maternal kin (Smith 2000) suggesting that groups not only split along matriline (Chepko-Sade & Stone Sade 1979, de Ruiter & Geffen 1998). Studies on baboons also found that social groups will be genetically structured by age as age cohorts tend to be paternal half-siblings (Altmann et al. 1996).

And finally, in species where males stay in their natal group with their kin, e.g., as in chimpanzees, is the picture of paternal kin bias reverse as in species, like the rhesus macaque, where females stay in their natal group together with their kin?