Letter to Editor

Some comments on the paper of Grant (2007)

Trivers and Willard (1973) (henceforth TW) noted provisions under which individual females could increase their expected number of grandchildren. The provisions were as follows:

• a.

  males have a higher variance of reproductive success than females,

• b.

  reproductive success correlates positively with female ‘condition’, and

• c.

  female ‘condition’ correlates positively with offspring ‘condition’.

TW noted that if these provisions were fulfilled, then an individual female in good ‘condition’ would maximise her expected reproductive success if she produced sons, whereas one in poor condition would do so by producing daughters. (The principal behind TW holds regardless of the sex having greater reproductive variance. However, since that sex is usually the male in mammals, I have chosen to formulate TW as above.) TW hypothesised that if these provisions were fulfilled, such skewed sex ratios actually exist at birth. In many mammalian species (including human beings), it is fairly well established that these provisions are fulfilled. For instance, men's reproductive variance has been shown to be greater than women's in such diverse contexts as the Yanomamo Indians (Chagnon et al., 1979), Venezuelan marriages (Chacon-Puignau and Jaffe, 1996), a large sample of deceased Australian men and women a century ago (Heron, 1914), and in contemporary North America (Low, 1988; Gaulin and Robbins, 1991). Yet though subsequent testing of TW has shown substantial success in some taxa, it has proved less satisfactory in others e.g. primates (West and Sheldon, 2002). For instance, only about 50% of the 54 studies on human beings (Lazarus, 2002) and 35 studies on non-human primates (Brown and Silk, 2002) supported TW at conventional levels of statistical significance. Of the other 50%, some produced results that were non-significant, but others produced results that were significantly contrary to TW. For this reason, Brown and Silk (2002) concluded that (bearing publication bias in mind), it was reasonable (provisionally at least) to accept the null hypothesis that adaptive TW variation in sex ratio does not exist in relation to maternal dominance. However, there are three points to make here:

• 1.

  More recent reviews by Sheldon and West (2004) and Cameron (2004) showed higher consistency with TW when the measure of maternal condition is considered as well as its timing.

• 2.

  Care should be taken not to assess TW on species for which the TW assumptions do not apply (Hewison and Gaillard, 1999).

• 3.

  If one were to allow that TW should apply to male as well as female parents, then there is a great deal more evidence in support of TW than that considered by Brown and Silk (2002). This evidence consists in the studies on human sex ratio variation by paternal occupational, medical and chemical exposures (e.g. as cited in James, 2006b). That paper summarised four forms of adverse occupational exposure, eight forms of adverse chemical exposure, and four forms of illness in association with which men (in conformity with TW) reportedly sire excesses of daughters. Thus, at first sight, there is too much evidence favouring TW to allow us to reject it, but still perhaps too much against it to allow us to accept it. The question arises whether the TW hypothesis is false or whether the operation of adaptive processes may be obscured (or negated) by constraints. The present note will consider Grant's response to this problem.

There is evidence that some mammalian sex ratio variation already exists at the time of conception (James, 2006a). Moreover, there is evidence that part of this variation is associated with the hormone levels of both parents around the time of conception, high levels of testosterone being associated with the subsequent births of sons (James, 1996, James, 2004). Grant (1998) noted that stress (as e.g. caused by captivity or food deprivation) causes mammalian females to produce high levels of adrenal androgens, and thus ex hypothesi and contrary to TW, a high proportion of sons. Grant (2007), following her earlier work (Grant, 1998, Grant, 2003), provided valuable and persuasive confirmation for this suggestion by making a number of observations.

First, she noted that it would resolve the apparent conflict between the observations of Verme (1969) and Clutton-Brock et al. (1984) that female deer in (respectively) poor and good condition reportedly both produced significant excesses of male offspring. At first sight, only the data of Clutton-Brock support TW. However, Grant's suggestion is that the anomalous male excess among the offspring of Verme's (1969) does was a consequence of high maternal adrenal androgens, whereas the male excess among the offspring of Clutton-Brock's dominant females was a consequence of dominance-associated high androgens. Grant (2007) strengthened the suggestion by citing further data confirming both studies.

Second, Grant (2007) amplified the power and scope of this form of explanation by noting its apparent potential to accommodate anomalies in the data relating to the local resource competition hypothesis of Clark (1978).

Grant (2007) offered plausible suggestions concerning the associations between captivity, dominance, and stress. Using these, she was able to reconcile of a number of (seemingly negative) studies of offspring sex ratios of captive mammals with adaptive sex ratio theory (e.g. Silk et al., 1981; Simpson and Simpson, 1982).

In short, building on her previous work, the suggestions in Grant (2007) seem to represent a major advance in the interpretation of data on mammalian sex ratios. They suggest (though the suggestion was not made by Grant) that apparent failures to confirm TW may (in many cases) be interpreted as a consequence of specifiable hormonally mediated constraints. Nevertheless, in the rest of this note I shall draw attention to what I take to be weaknesses in this latest presentation of what she still calls her ‘present maternal dominance hypothesis’ (her Section 2.1). The following two questions will be asked:

• 1.

  Are females the sole source of mammalian sex ratio variation, as Grant urges?

• 2.

  Which hypothesis do Grant's data and argument support: that a major proximate determinant of sex ratio variation is maternal dominance or that it is maternal testosterone?