Serial monogamy as polygyny or polyandry? : marriage in the tanzanian pimbwe

Abstract

Applications of sexual selection theory to humans lead us to expect that because of mammalian sex differences in obligate parental investment there will be gender differences in fitness variances, and males will benefit more than females from multiple mates. Recent theoretical work in behavioral ecology suggests reality is more complex. In this paper, focused on humans, predictions are derived from conventional parental investment theory regarding expected outcomes associated with serial monogamy and are tested with new data from a postreproductive cohort of men and women in a primarily horticultural population in western Tanzania (Pimbwe). Several predictions derived from the view that serial monogamy is a reproductive strategy from which males benefit are not supported. Furthermore, Pimbwe women are the primary beneficiaries of multiple marriages. The implications for applications of sexual selection theory to humans are discussed, in particular the fact that in some populations women lead sexual and reproductive lives that are very different from those derived from a simple Bateman-Trivers model.

Fig. 1

Variance in (a) fertility and (b) reproductive success for men and women who have reached their forty-fifth birthday. The sex difference in variance in fertility is significant, but not the sex difference in reproductive success (see text). Comparing women with men who have reached their fifty-fifth birthday (n = 87) shows no significant differences in variances for fertility or reproductive success (see text)

Fig. 2

Survival analysis of the last intermarriage interval (in years) for Pimbwe men and women. Cox’s regression shows significant effects of sex (Wald = 5.64, p = 0.018), age (Wald = 5.87, p = 0.015), and an age × sex interaction (Wald = 8.30, p = 0.004)

Fig. 3

The associations between number of spouses and (a) fertility and (b) reproductive success for men and women. The mean is shown with a circle, and the standard error (*2) with a bar. For statistics see Table 2

Fig. 4

Work ethic for men and women according to number of spouses. Chi-square linear by linear association is significant for men (11.16, df = 1, p = 0.001) and women (3.82, df = 1, p = 0.050). Statistical analyses dropped individuals with no spouses

Fig. 5

Alcohol consumption for men and women according to number of spouses. Chi-square linear by linear association is marginally significant for men (3.10, df = 1, p = 0.078) but not for women (2.06, df = 1, ns). Statistical analyses dropped individuals with no spouses