Ethnic kinship


Quantification and Increase through Political Evolution

Frank Salter

[Original details:]

Max Planck Society, Human Ethology,

Von-der-Tann-Str. 3,

82346 Andechs, Germany

Notes for a report presented at the 16th biennial meeting of the International Society for Human Ethology, 7-10 August 2002, Montreal


Biological theories of the origin of heroism in warfare and other types of altruism directed towards the tribe or ethnic group have often attributed this to some adaptive function, such as retention of group resources. However, without an estimate of the aggregate kinship at stake within the group, no theory of altruism can be tested using W. D. Hamilton’s rule for adaptive altruism. By “adaptive”, Hamilton meant evolutionarily stable, such that the altruist’s genes are not selected out of the gene pool. Though Hamilton’s 1975 model showed that ethnic kinship could theoretically be large, no evolutionary theory has yet answered the most basic question, whether in fact ethnic kinship—the genetic similarity of co-ethnics who are not genealogical kin—is ever large enough to make ethnic altruism adaptive.

Harpending (2002) derived a population-genetic formula for estimating the aggregate ethnic kinship held by one population in relation to another based on the genetic distance between the two populations. The genetic assay data needed to make this estimate for modern ethnic groups are becoming available. The data used in this present study are provided by Cavalli-Sforza et al. (1994). Based on those data, aggregate ethnic kinship is much larger than aggregate family kinship. Data on tribal genetic distance are uncertain. But existing evidence indicates that tribal genetic interests vis a vis neighbouring tribes in the Neolithic were already larger than familial kinship. The direction of theory and data strongly indicate that self sacrificial altruism in warfare could have been adaptive from that time.

The Hamilton-Harpending algorithm offers an analytical tool for estimating whether a population was (or is) a fit object for altruism, and thus whether that altruism was (or is) sustainable across evolutionary time.


Hamilton’s Fst statement referred to genes coding for altruism, not to the whole genome. However, my point regarding kinship remains valid because I used Fst data based on sampling of the genome, not on altruistic genes.

Also please note that the more accurate data provided by the Human Genome Data Base show somewhat lower racial variation and therefore lower racial kinship. Instead of 9%, the French-Japanese variation is 6% (Salter and Harpending 2013). Because the reduction is not great it does not invalidate the analysis.

Salter, F. K., & Harpending, H. (2013). J. P. Rushton’s theory of ethnic nepotism. Personality and Individual Differences, 55, 256-260. doi:10.1016/j.paid.2012.11.014

I also should correct one part of my summary of David Goetze’s insight concerning collective goods, such as big-game hunting and collective defence. From about the 3 minute mark I say that these cooperative activities allow large investments to be made in large populations. Actually, they also allow small investments to make a difference.


The key issue in the evolutionary theory of ethnic conflict is whether solidarity towards fellow ethnics has been adaptive. Components of this problem are:

(1) Was the kinship between random members of bands and tribes large enough for altruism directed towards fellow ethnics to have been adaptive?

(2) If the answer to (1) is yes, then what mechanisms were necessary? Answering this question will help locate the stage in political evolution at which ethnic altruism could have become adaptive, thus allowing genes or culture that code for ethnic altruism to spread through the population.

We already know the answer, or much of the answer, to the second question. Proponents of group selection have argued, convincingly I think, that members of bands and tribes can behave altruistically without being selected out by free riders. Eibl-Eibesfeldt (1982) argued from his field observations that mutual monitoring, ubiquitous in small-scale societies, is sufficient to suppress cheating. He pointed to the pronounced group identity and mutual support found in primitive societies, and argued that this originated in kinship bonds. The cohesion of band and tribal societies makes them units of selection, Eibl argued. This point was elaborated by Boyd and Richerson (1992), who argue that monitoring and punishment are so effective in small scale societies that they allow the evolution of cooperation, or any other characteristic that is promoted by a culturally-governed group strategy.

Whether or not one accept that group selection has figured in human evolution, the mechanisms advanced by group selectionists are sufficient to allow a more conservative process, extended kin selection, to occur. In fact this is what Eibl has always meant by his version of group selection.

A final mechanism deserving of mention is collective goods. A criticism of extended kin selection is that it is impossible for an individual effectively to invest in a kin group much larger than a family, because the benefit would be spread so thinly that the payoff would always be greater from investing in close kin, rather than distant ones. Goetze (1998) has dispelled this concern. He draws on economic theory to argue that by contributing to collective goods—such as hunting large game animals or defending the group—allows an individual to confer a large fitness benefit on a large number of individuals.

So there is no mechanical problem with the feasibility of individuals showing altruism to kin groups larger than the extended family. Indeed, all these mechanism—control of free-riders, bonding the group, and choosing or fashioning collective goods—are highly scalable. They can be increased in scale to accommodate a kin group of any size. Admittedly some novel and ingenious social devices are needed to perform these functions for large groups, but humans are ingenious, as is clear from the many experiments in political evolution.

Thus the second problem in understand the evolution of ethnocentrism the second is already solved, or well on its way to being solved. It’s the first problem that remains; indeed, it has hardly been addressed. To reiterate, was the kinship between random members of bands and tribes large enough for altruism directed between them to have been adaptive?

The question should be recast in light of Goetze’s analysis of collective goods. I shall use the term ‘patriotism’ to mean altruism towards a collective good. When collective goods are available to which individuals can contribute, is the aggregate kinship of the whole group sufficiently high to allow patriotism to be adaptive, i.e. evolutionarily stable? Dawkins thinks not. He maintains that only altruism shown to close kin is adaptive. But Hamilton disagreed. In his classic 1975 paper, ‘Innate social aptitudes of man: An approach from evolutionary genetics’, he discarded the notion that inclusive fitness processes can only operate between genealogical kin, and argued that altruism can be adaptive between anonymous, genetically similar individuals.

“[C]onnections which the remote townsman does not so easily know of make up in multiplicity what they lack in close degree” (1975, p. 142).

By townsman Hamilton means the member of a band or tribe. He showed mathematically that even with a steady trickle of migration between populations, relatedness can rise as high as 0.5 between random members. Hamilton concluded that altruism on behalf of the group could then be adaptive, especially if it preserved the group from replacement. The point that inclusive fitness processes can operate between individuals merely on the basis of genetic similarity, without any genealogical information, is critical, and I quote Hamilton’s commentary on this theoretical advance.

“Because of the way it was first explained [by Hamilton], the approach using inclusive fitness has often been identified with “kin selection” and presented strictly as an alternative to “group selection” as a way of establishing altruistic social behaviour by natural selection. But…kinship should be considered just one way of getting positive regression of genotype in the recipient, and that it is this positive regression that is vitally necessary for altruism. Thus the inclusive fitness concept is more general than “kin selection” ” (Hamilton 1975, pp. 140-41; [p. 337 in the 1996 reprint]).

This frees the analyst from the “identical by descent” clause in Hamilton’s original (1964) formulation, allowing the direct measurement of kinship processes using genetic assay data. These data are usually expressed not in terms of kinship coefficients, but genetic variation, for example FST. However, Harpending (1979) provides a formula for converting FST measures to kinship coefficients.

fo = FST + (1 – FST)[ – 1/(2N – 1)]

where fo is the local kinship coefficient, FST the variance of the metapopulation, and N the overall population. Within primordial dialect groups and tribes, where N is approximately 500, the second complex term in this equation is small. When N is large, as it usually is with modern ethnies, a good approximation for the above equation becomes, simply:


(The kinship concept needs clarification. In population genetics the coefficient of kinship, f, between two individuals is defined as the probability that an allele taken randomly from one will be identical to an allele taken at the same locus from another. This definition is close to that of Hamilton’s (1964) original coefficient of relatedness r, which he used in his classic formulation of inclusive-fitness theory, except that in simple cases 2f = r. This means that parental kinship is 0.25, not 0.5. Kinship to self is 0.5, not the familiar 1.0, which refers to relatedness r. A fuller explanation is provided in Salter [in press])

Harpending’s simple formula allows the estimation of average kinship within local populations based on FST measures. The principle can be simply stated thus: variation between two populations is equal to kinship within each of them. As a hypothetical example, if the variation between two groups P and Q is FST = 0.25, then the kinship between two randomly-chosen members of P is likewise 0.25, or that of sibs or parent and child. The same applies to random pairs drawn from Q.

This brings us to the subject of this presentation: Was there sufficient genetic variation between primordial human groups for individual inclusive fitness to be boosted by acts of ethnic solidarity, by patriotism?

Let’s begin with the band, numbering between 30 and 50 individuals, comprised of two or three extended families connected by marriage ties. I could not locate data on inter-band genetic variation, but Harpending (personal communication) reports that inter-band FST is typically small, 0.01 or less. Let us assume, for illustrative purposes, that it is 0.0005. If, apart from extended family, a band numbered, say, 25 individuals, then this group’s aggregate kinship to a random individual is 0.0005 x 25 = 0.0025, which is the equivalent of one hundredth of a child. This number only has meaning in the context of competition with a neighbouring band. It will be much higher in the context of competition with more genetically distant populations. By comparison to this vanishingly small kinship, an individual’s genealogical kin might represent the genetic equivalent of five or six children (3 actual children plus cousins, grandchildren, etc.). The selection advantage of altruism towards nonkin would usually be outweighed by altruism towards kin. Nevertheless, band solidarity might have paid off because the fate of the extended family was inseparably bound up with the fate of the band. The average kinship with the band would have been high relative to the average kinship with members of neighbouring bands. (An approximation: assume that family plus others yield the equivalent of six children within the band, or an aggregate kinship of 1.5. Then average kinship is 1.5/50 = 0.03. Average kinship with neighbouring bands is –0.01.)

Genetic variation grows with the geographic scale of population units, so that dialect and tribal populations have higher kinship between random pairs than do bands. Typical variation between small dialect groups and tribes might be 0.005. FST between clusters of Bantu tribes is much higher, typically about 0.015. Between West African populations Fst varies from 0.0013 (Ewe-Volta) to 0.049 (Volta-Wolof). The average is about 0.02 (Cavalli-Sforza et al. 1994, p. 181). Neighbouring American Indian tribes have a typical genetic distance of about 0.025 (Cavalli-Sforza et al. 1994, p. 323). The Americas show high genetic variability, with an average FST of 0.070, compared to Australia’s 0.019, Polynesia’s 0.031, New Guinea’s 0.039, sub-Saharan Africa’s 0.035, and Caucasoid’s as a whole of 0.043 (Cavalli-Sforza et al. 1994, p. 336).

Genetic variation continues to increase with geographical , though recall that we are discussing autochthonous populations, those that have been resident in an area for many thousands of years. Cavalli-Sforza et al (1994, p. 122) have charted the relationship between FST and distance within large regions.



Fig. 1  The relationship between genetic distance and geographic distance within continents. Note that the curves are based on pre-colonial populations (from Cavalli-Sforza et al. 1994, p. 122).


Between continents genetic variation increases greatly. Table 1 shows the FST distances between geographical races, which can be characterized as continental-scale populations.

Africans 0.0
Non-European Caucasoids 1340 0.0
European Caucasoids 1656 155 0.0
Northeast Asians 1979 640 938 0.0
Arctic Northeast Asians 2009 708 747 460 0.0
Amerindians 2261 956 1038 747 577 0.0
Southeast Asians 2206 940 1240 631 1039 1342 0.0
Pacific Islanders 2505 954 1345 724 1181 1741 437 0.0
New Guineans and Australians 2472 1179 1346 734 1013 1458 1238 809


Table 1. Genetic variation between nine geographical races, measured as FST x 10,000 (From Cavalli-Sforza et al., 1994, p. 80; rounded to nearest integers; standard errors omitted).

Inter-racial variation is typically as high as 0.125 or even 0.25 (between Pacific Islanders and Africans). In the latter case, intra-racial kinship is the equivalent of parental kinship. Higher variation across greater geographical distances means that populations organized competitively over those distances have higher within-population kinship. At the same time, aggregate kinship will increase due to the larger size of the polity. In other words, other factors being equal, group solidarity becomes more adaptive as the scale of political organization grows. In Table 2 I estimate the aggregate kinship in child-equivalents for different types of populations. The values differ for each continent, but the FST values adopted are realistic.

Child equivalents
N Inter-pop. FST Extended family kinship Non-family group members
Band 50 0.0005 5
Dialect group 500 0.005 5 10
Large tribe 5000 0.01 5 200
Modern nation 10 mill. 0.015 5 600,000
Racially different nations 10 mill. 0.125 5 5 mill.

Table 2.  Distribution of aggregate kinship in different sized autochthonous populations based on genetic distance to neighbouring populations of the same kind.


Table 2 indicates that beyond the band, ethnic solidarity could have been adaptive, assuming that competition existed between the larger social units, that free riders were controlled and that collective goods existed in which to invest.

From the emergence of tribes in the Neolithic, social organization spanning many miles would have created scope for collective goods that benefited many hundred or thousands of individuals. The positive relationship between geographic and genetic distance would have created an adaptive opportunity for aggressively expansive group strategies, perhaps in the autocatalytic process postulated by E. O. Wilson:

“A band might then dispose of a neighboring band, appropriate its territory, and increase its own genetic representation in the metapopulation, retaining the tribal memory of this successful episode, repeating it, increasing the geographical range of its occurrence, and quickly spreading its influence still further in the metapopulation. Such primitive cultural capacity would be permitted by the possession of certain genes” (E. O. Wilson 1975, p. 573).

Eibl-Eibesfeldt (1982) makes essentially the same point, by emphasizing group cohesion and territorial displacement. Likewise Hamilton combined the factors of aggressive territorial expansion.

“[P]rimate examples suggest the prototype war party as an all-male group, brothers and kin, practised as a team in successful hunting and at last redirecting its skill towards usurping the females or territory of another group. Out of such cells can be built the somewhat less stable organism of the postneolithic army. . . . If the male war party has been adaptive for as long as is surmised here, it is hardly surprising that a similar grouping often reappears spontaneously even in circumstances where its present adaptive value is low or negative, as in modern teenage gangs.” (Hamilton 1975, p. 148)

The key elements in the strategy would have been capturing territory and replacing the conquered population in whole or part. Ethnic nepotism in the form of advancing such a strategy or defending against it would have yielded fitness payoffs much larger, though less regularly, than familial nepotism.

The Hamilton-Harpending algorithm offers an analytical tool for estimating whether a population was (or is) a fit object for altruism, and thus whether that altruism was (or is) sustainable across evolutionary time.

Combining inclusive fitness theory with gene assay data has implications for the debate regarding group selection of altruism directed towards ethnies. Research attention long focused on the possibilities of group selection of altruism should be widened to look for the preconditions for extended kin selection: ethnic kinship; control of free riders; and the availability of collective goods facilitating ethnic continuity.


 Boyd, R. and Richerson, P. J. (1992). Punishment allows the evolution of cooperation (or anything else) in sizable groups. Ethology and Sociobiology, 13: 171-195.

Cavalli-Sforza, L. L., Menozzi, P. and Piazza, A. (1994).  The history and geography of human genes. Princeton University Press, Princeton, New Jersey.

Eibl-Eibesfeldt, I. (1982). Warfare, man’s indoctrinability and group selection. Ethology (Zeitschrift für Tierpsychologie), 60: 177-98.

Goetze, D. (1998). Evolution, mobility, and ethnic group formation. Politics and the Life Sciences, 17(1): 59-71.

Hamilton, W. D. (1964). The genetic evolution of social behavior, parts 1 and 2. Journal of Theoretical Biology, 7: 1-51.

Hamilton, W. D. (1975). Innate social aptitudes of man: An approach from evolutionary genetics. In Biosocial anthropology, (ed. R. Fox), pp. 133-55. Malaby Press, London.

Harpending, H. (1979). The population genetics of interactions. American Naturalist, 113: 622-30.

Salter, F. K. (2002). Estimating ethnic genetic interests: Is it adaptive to resist replacement migration? Population and Environment, 24(2): 111-40.

Wilson, E. O. (1975).  Sociobiology: The new synthesis. Harvard University Press, Cambridge, MA.

Comments are closed.