If we look back at the predicted results according to each hypothesis we can see quite clearly that inbreeding avoidance is more strongly supported than intra-sexual competition by the results of the correlation analysis. On re-examination of figure one we can see that a strong positive relationship exists between a females age at their first conception and the age at natal dispersal of the species. In the following discussion unless stated otherwise males refers to the dispersing sex. Figure two shows a similar relationship when one considerstenue length. This certainly appears to be evidence supporting inbreeding avoidance, yet we cannot fully rule out intra-sexual competition at this stage. Figure 3 however does seem to make it unlikely, in this we see positive relationships between dispersal age and the average numbers of adult males and females in a group. Tenure length has a similar pattern of relationships. The relationship to female numbers can be interpreted as support for both hypotheses. An increased number of adult, and thus breeding, females represents an increased resource base, thus lessening the costs of competition and allowing longer residence. However, concomitant with a high number of females is a high number of unrelated females. Unrelated sex ratio has been reported by Paul and Kuester (1999) and by Kuester et al (1994) as a key factor in the decision to emigrate. In essence both come down to the same thing: dilution of costs. Ambiguous though the female data may be the male data is anything but. The expectations of the two hypotheses call for diametrically opposed relationships. Intra-sexual competition predicts an negative relationship, as male numbers rise thus residence time falls due to increasing competitive pressure. A positive relationship, on the other hand, is what one would expect if inbreeding avoidance is behind the timing of dispersal. Increasing male numbers will mean increased female numbers. Hand in hand with this is the aforementioned rise in unrelated sex ratio, thus the impetus to disperse is reduced. However, correlation and regression analysis revealed the relationship between tenure length and male numbers to be insignificant. Side from these predicted relationships others were found. Inter-birth interval was found to have a strong positive correlation to both dispersal age and to tenure length whilst lifespan was positively related to dispersal age only. These relationships at first glance seem somewhat anomalous, however the role of lifespan can easily be explained away. The longer-lived a species is, the later an individual of that species will disperse from it’s natal group. Given that increased longevity results in longer ontogeny (Stephens 1980) one can argue that if, as Smuts (1985) and Rasmussen (1981) have hypothesised, there is an anthropometric threshold to dispersal this will in turn be reached at a later date. Inter-birth interval on the other hand is a more ambiguous factor, one can see that logically a longer lifespan will entail a longer inter-birth interval yet it appears that there is more to it than simply a question of allometrics. However, during partial correlation analysis correcting for differences in inter-birth interval rendered all relationships to and between dispersal age and tenure length non-significant whilst doing the same for lifespan did not. This implies that the influence of inter-birth interval goes beyond simply being a correlate of lifespan. Indeed when differences in lifespan were corrected for inter-birth interval retained a significant, positive relationship to both dispersal age and tenure length. Indeed linear regression analysis revealed that lifespan, far from being a significant predictor of inter-birth interval actually never explains more than 4% of the variance. Inter-birth interval seems to be more related to factors such as female age at first conception and group size.

In comparing female and male dispersing species one can see intriguing patterns emerge. If we first deal with dispersal age one can see that amongst the male dispersing species the predictions of inbreeding avoidance seem to be fulfilled. The largest r² value and greatest significance are attached to female age at first conception. If we consider the other two contributing factors, tenure length and infanticidality a more subtle influence appears. Infanticidality has an r² of 0.175 while tenure length has a slightly lower value of 0.101. These two variables can be seen as inter-related. In infanticidal species sub-adult males are forced out of the group by the new a male (Struhsaker and Leland, 1987). Under these circumstances we can see that infanticidality of a species will logically affect the timing of dispersal. The potential costs of inbreeding are less than those of being dead. Thus the tenure of an infanticidal male will determine the timing of dispersal of his sons.

Moving to the female dispersing species dispersal age is predicted solely by lifespan and inter-birth interval. Given that this is based solely on four data points one should perhaps take these findings with a pinch of salt however one can suggest possible explanations. Three of the four species are very long-lived hominoids, it may well be that lifespan is related as a scaling element and that inter-birth interval, which is related to group size and female age at first conception is acting to mask the influence of these variables. Tenure length, compared across the two dispersal patterns, shows two distinct forms. Where males disperse tenure is determined by group size and its correlate of the number of adult males. Here again we see support for the inbreeding avoidance hypothesis, the relationships between tenure length, group size and male numbers are strongly positive, in keeping with the predictions of the hypothesis. Where males remain philopatric however the male age at sexual maturity is the key determinant of tenure length, having a massive r² of 0.92. Again the smallness of the data set makes ascribing cause and effect spurious at best it certainly seems to lend support to inbreeding avoidance. The residence length of one sex not exceeding the age at sexual maturity of the other as is illustrated in figure 5.

As previously mentioned attempts to sub-divide the data set into its constituent genera for examination failed in all but the Macaca species. Here again the predictions of inbreeding avoidance are fulfilled. Dispersal age is explained solely by female age at first conception, initially this seems heartening but the r² is merely 0.34. This leaves 66% of the variance in dispersal age unexplained, yet if it is not due to any of the other recorded variables then what can explain it? Numerous possible explanations present themselves, Paul (1999) suggests that the seasonality or otherwise of breeding may affect the timing of dispersal in Macaca species. Other factors include the history of the group, in as much as a group that has recently split from another group will tend to have a higher relatedness (Chepko-Sade and Olivier, 1993) thus giving a greater impetus to dispersal. Tenure length seems unusual at first glance, group size is the sole predictor, having a relatively low r² value (0.316). This certainly fulfils the expectations of inbreeding avoidance, as we can see from figure 6 the relationship is positive. Indeed further examination shows that group size is strongly related to female age at sexual maturity, thus lending another level of support to the inbreeding avoidance camp.

Moving to the subdivision according to mating system the age at dispersal and indeed the tenure length of species with a multi-mount ejaculation system can be predicted quite accurately based solely on the genera of the species concerned. In the case of dispersal age the only other variable explaining more than 3% of dispersal age variance is tenure length. Interestingly the relationship between tenure length and dispersal age is a negative one, as shown in figure 6. The longer a male is resident in a group, the sooner his sons will disperse from that group. This is an ambiguous finding and does not fit with either theory particularly well. One could suggest that a longer residence time would entail larger numbers of offspring, thus more possible brother-sister inbreeding which the males would avoid by leaving as soon as possible. Alternatively one can argue that any male who resides for a long time in a group must posses superlative competitive skills, thus his sons are at an increased disadvantage when it comes to competing, be it for mates or for food and will leave sooner as a result. The relationship between dispersal age and tenure length is a reciprocal one, 41.3% of the variance in tenure length can be explained by dispersal age variation. This suggests that a more subtle range of influences are acting on the multi-mount ejaculators than just intra-sexual competition and inbreeding avoidance.

Moving to the single-mount ejaculators dispersal age certainly seems to be determined by inbreeding avoidance, female age at first conception has the huge r² value of 0.914 and again the relationship is positive (Figure 7). This is a logical relationship given the nature of single mount ejaculation mating, couplings are very brief, lasting a matter of seconds (Coulson, 1996). The relevance of this is that an incestuous mating is more likely to be successful in that there is less time to be interrupted than in the multi-mount system. Thus one can argue that selection appears to have allowed little variation in dispersal timing. As regards tenure length, of the six variables that constitute the predictive formula five have a collective r² of a mere 0.63 compared to the 0.93 accorded male age at sexual maturity. This is anomalous, one would expect female age at sexual maturity or first conception to be related if inbreeding avoidance was the key or some measure of group size if intra-sexual competition was the ultimate cause. Examination of the influences on male age at sexual maturity in the hope of revealing a masked relationship to one of the above mentioned factors reveals nothing of the kind. Inter-birth interval is the main predictor at an r² of 0.941. Curiously examining inter-birth interval reveals that male age at sexual maturity is its key predictor, with an identical r² of 0.941. Despite this somewhat unusual set of relationships one can re-examine correlation analysis and discern a logical relationship. Here again we see tenure as related to male age at sexual maturity, yet if we examine the variables relating to it we see that the strongest relationship is to female age at first conception with an r of 0.92 and a significance level of less than 0.001. Quite why this relationship did not appear during regression analysis is unclear, however both male age at sexual maturity and female age at first conception are very well predicted by inter-birth interval, with r²’s above 0.9. Thus perhaps there is a more complex interplay between these variables that is more subtle than the analysis could discern.

The infanticidal species reveal a problematic combination of influences. Dispersal age is determined by female age at first conception and at sexual maturity . This is a clear endorsement of inbreeding avoidance yet it does not allow for the nature of the species concerned. One would expect that dispersal age would be related positively to tenure length, in that, as previously stated, new a males will expel or kill sub-adult males (Struhsaker and Leland, 1987). This picture is not seen however and it is difficult to see why not. Accepting that relationships within a multi-male multi-female group will be less clear cut than those of a uni-male group, such as the widely studied Hanuman langurs (Presbytis entellus) one would still expect something to be seen. Examination of the influential variables in the search for secondary influences revealed no role for tenure. It may possibly be therefore that within a multi-male multi-female group sub-adult males are able to avoid the risk of infanticide somehow, perhaps via protection by maternal relatives. If this form of social protection occurs one can suppose that the cost of inbreeding becomes paramount to avoid once the cost of being dead is removed. Considering tenure length we again see the importance of inbreeding avoidance. Female age at sexual maturity being the key ingredient of the derived formula. Again this is unexpected, one would predict that the tenure of an infanticidal male would be related to either some measure of his social network and competitors ,such as male and female numbers, or some intangible variable such as the presence of extra-group males, age and so on. Male numbers do play a part, but a very small one, having an r² of a mere 0.009. Here again we can presume that some other untested variable is playing a role as testing for secondary influences revealed nothing that even hinted at what one would expect.

In the case of the non-infanticidal species dispersal timing is, for the most part, determined by male age at sexual maturity, with an r² of 0.937. This certainly points to inbreeding avoidance as the ultimate cause of dispersal. Indeed the huge importance of male age at sexual maturity to these species dispersal hints at a heavy cost to inbreeding. Smith (1995a) demonstrated that the degree of consanguinity needed before inbreeding depression will be seen varies between species and indeed Moore and Ali (1984) suggested that it varies between populations. It is an intriguing thought to consider that infanticide, increasing infant mortality may in fact act in much the same way that Ralls and Ballou (1988) argue inbreeding depression does. This is to say it acts to remove deleterious alleles thus necessitating a closer relationship before inbreeding depression is seen. If, as previously suggested, sub-adult males in a multi-male group avoid infanticide via protection from maternal relatives one can suggest that infanticide acts to weed out those with alleles not deleterious enough to kill them in infancy, when congenital defects are most likely to appear (Packer, 1979). Jones (1995) having previously shown that inbred mothers mistreat their offspring in M.fascicularis perhaps their protective skills are not as great? Perhaps their offspring are slower or weaker (Wildt et al 1982) and thus fall victim to infanticide more easily. Therefore in non-infanticidal species we can extend this argument to suggest that incestuous matings are riskier as this secondary filter is absent, thus giving males a greater incentive to depart as soon as they become sexually mature. The tenure of non-infanticidal males is again seemingly determined by inbreeding avoidance, female age at sexual maturity and at first conception have a combined r² of 0.766 and as expected the relationship is positive.

It is quite clear by now that regardless of how the data set is sub-divided inbreeding avoidance is the ultimate cause of natal and secondary dispersal. Intra-sexual competition falls flat in comparison, one can perhaps suggest a proximate role in determining the precise date of dispersal or the group emigrated to but in terms of ultimate influence it is impotent.

If we now consider why the dispersing sex would delay their dispersal, something that increases the risk of inbreeding. This risk is illustrated in figure 9 which shows that the latest age at natal dispersal exceeds both male sexual maturity and female first conception, leaving a window of opportunity for inbreeding. Given this why run the known risks? Correlation analysis revealed certain relationships, without correcting for anything no relationships were found. However once partial correlations were employed this soon changed. The three key variables appear to be inter-birth interval, lifespan and male age at sexual maturity. When one corrected for differences in these three variables all other relationship became non-significant. As mentioned previously we can propose that lifespan is acting as a scaling element, the longer the lifespan the later the dispersal. This suggestion does not hold true for inter-birth interval and male age at sexual maturity, regression analysis showed lifespan to be unrelated to male age at sexual maturity and only minimally so to inter-birth interval. The strongest relationships to inter-birth interval and male age at sexual maturity were between oneanother, having r² of greater than 0.9. It is this relationship that can allow us an explanation. Given that dispersal is strongly related to male age at sexual maturity, logically, accepting that what we are looking at is one extreme of a distribution we must see that as male age at sexual maturity increases so will the age at dispersal and thus the range will shift upwards too. Inter-birth interval seems to be dragged along. Regression analysis of the whole data set revealed inter-birth interval to be the key variable, this could be masking male age at sexual maturity given the strong relationship between the two detailed above. Additionally it could be that a longer inter-birth interval allows a longer time before inter-generational inbreeding can occur, allowing longer residence. If one examines the sub-divided data set interesting relationships appear. In male dispersing species no predictors are found, suggesting that dispersal is delayed due to some variable not included in this study. Where females disperse female age at first conception is the key, this seems to be the same role as male age at sexual maturity plays in the dataset as a whole. What is interesting is that amongst the baboons the only predictor found is tenure length. This confirms a relationship found by Green (1984). The relationship he found was that male baboons could choose different strategies to emigration, one could go as a sexually mature yet not full grown individual and join a large group and stay for a long time or, alternatively, wait until fully grown and acquire a smaller group of ones own which would be held for a shorter period.

This implies that the later one leaves the natal group the shorter your tenure will be. Moving to the Macaca species we see that female age at sexual maturity is the key thing, again seemingly simply altering the scale for dispersal. We can see that, with the possible exception of the baboon species a delayed dispersal is simply a result of avoidance of inbreeding.

One can now consider what makes males disperse in some species and females in others, what are the essential differences that cause a switch in the sex roles? With an odds ratio of 90.48% inter-birth interval was revealed as the key factor. If we accept that an increased inter-birth interval as a logical consequence of the far greater lifespan of three of the four female dispersing species (Pan paniscus, Pan troglodytes, Gorilla gorilla berengei) we can move towards an explanation for why females disperse. Wrangham (1980) argues that male dispersal occurs as a mechanism of inbreeding avoidance when females control food resources. The emigration of females suggests that inbreeding is costlier to them compared to males, given the disparity in reproductive potential between the two sexes (Trivers 1972) this is logically so. Stephens (1990) contends that among the hominoids the longevity of males results in a shift in the sex bias of dispersal. This contention is based on the fact that, in shorter lived species, male natal emigration avoids mother-son and brother-sister inbreeding. The short tenures of males, due either to rapid replacement or death avoids father-daughter matings. In the hominoids the increased longevity and more social forms of conflict makes this mechanism less likely. Studies in humans (Shepher 1983) and in P.troglodytes (Goodall 1986) have shown father-daughter matings to be far more common than mother-son. Physically the disparity in size is similar but the mother-offspring bond seems to restrict sons behaviour to their mothers. If we accept that females have higher costs to inbreeding than males, are less able to physically resist incestuous matings and that lifespan keeps fathers and daughters in the same group as the daughters mature then one can see how females come to disperse. This still leaves the question of why C.badius females disperse, their lifespan is shorter than the hominoids. Moore (1984) presents an argument that the leaf eating colobines should be an exception to Wrangham’s (1980) model. Clutton-Brock (1989) shows that in polygynous species with female dispersal the males have unusually long tenures, exceeding the age at which daughters become sexually mature. Thus it is that females disperse from the red colobus groups.

If we now examine the implications of these findings we will see the following:

1) Natal and secondary dispersal are primarily related to inbreeding avoidance, regardless of the sub-division of the data.

2) In all but the baboon sub-group late natal dispersal was related to inbreeding avoidance. In the baboon species it may represent a reproductive strategy choice.

3) Female versus male dispersal seems to be a matter of longevity, as proposed by Stephens (1980)

Looking across species we can see a definite consistency. Mammalian species are predominantly male dispersing (Greenwood 1980). Birds, on the other hand are female dispersers. Studies have pointed to inbreeding avoidance as; Pusey and Packer 1987, Packer and Pusey 1993) the ultimate cause of dispersal in several species, such as ; Horses (Equus caballus; Monard et al 1996; Rutberg and Keiper, 1993); White footed mice (Peromyscus leocopus; Wolff, 1992); Red howler monkeys (Alouatta seniculus; Agoramoorthy and Rudran 1993); the Easter grey Kangaroo (Macropus giganteus; Jarman and Southwell, 1986); African lions (Panthera leo; Packer and Pusey 1993, Pusey and Packer 1987).

The available evidence certainly suggests that inbreeding avoidance is the ultimate cause of dispersal. Yet, if we consider Stephens’ (1980) theory that inbreeding avoidance via secondary dispersal is due to the tenure of the males being curtailed by either death or replacement we are left with an intriguing quandary. The results revealed herein do not support this at first, female age at sexual maturity and at first conception are the keys to tenure length, not measures of competition. Correlates of lifespan such as inter-birth interval often come into view yet correlates of competition, when found to be significant, have relationships to tenure length that are in the opposite direction to what one would expect according to Stephens (1980). Sapolsky (1998) suggests a scenario that may help to explain this. He found that, amongst baboons, adult males who relinquish their position as a voluntarily encounter less aggression from the new a individual. This jump before being pushed strategy could mean that adult males are aware of their declining powers and may step down before competition becomes a factor.

Clearly this discussion of inbreeding and its avoidance has touched on only a few of the many facets of a highly complex range of behaviours. Inevitably questions remain. If, for example, as has been suggested, groups exchange males non-randomly (Melnick and Pearl 1987; Meikle and Vessey 1981; Cheney and Seyfarth 1983) then should this not lead to inbreeding when a son emigrates to a group which his father was born into? Cheney and Seyfarth (1983) argue that as long as a proportion of secondary dispersal is random and over long distances inbreeding can be avoided. Tentative support for this can be found in the work of Rogers and Kidd (1996). Examining a P.cynocephalus population in Tanzania they found, via Restriction Fragment Length Polymorphism (RFLP) paternity analysis that two-thirds of infants were sired by males from neighbouring troops. The other third however came from males form further afield. This mixture of paternity fits with Cheney and Seyfarth’s (1983) argument and the baboon population was found to be outbred. Further questions remain as to the proximate causes of dispersal, does intra-sexual competition have a role here? If, as proposed by Henzi and Lucas (1980), emigration is a female mediated process, how do they mediate it? how are their decisions arrived at? Sadly the primate studies are rarely as wide ranging and elegant as those from other species, whilst one can argue that this is an inevitable consequence of habitat and location the work of Monard et al (1996) provides an excellent template for future studies. Working on a wild, non-provisioned yet habituated population of E. Caballus in the Camargue region of France they were able to combine genetic paternity analysis for every individual born in the thirty years since the colony’s foundation with daily behavioural surveys from the same period. With this data the authors were able to determine the proximate and ultimate causes of dispersal amongst these animals. In comparison primate studies will pick up on perhaps one or two elements of the whole i.e. attraction to novel females (Cheney and Seyfarth 1983), kin recognition (Paul and Kuester 1993).

This thesis has dealt with the purely statistical aspect of dispersal and found inbreeding avoidance to be the ultimate cause of dispersal, yet without systematic examination of the ethological aspects of emigration we cannot hope to lay to rest the debate over ultimate cause nor to fully comprehend the diversity of dispersal patterns seen.