Trivers-Willard and the Mighty Males Hypothesis
Trivers and Willard first proposed that sex allocation should be sensitive to the environment (Trivers & Willard 1973). In their verbal model, females produce more sons when sons are expected to become high-condition adults, where condition refers to general health, energy reserves, vigor, and overall quality. The logic is that male fitness is limited by mating opportunity, whereas female fitness is limited by gamete production (Bateman 1948). As a result, males experience stronger sexual selection (Trivers 1972; Singh & Punzalan 2018), and male reproductive success (number of successful inseminations) will be exponentially distributed with a modal value of zero (Jones et al. 2002) (Fig. 2a). In other words, a small proportion of males capture a large portion of the mating opportunities. Further, male phenotype is non-random with respect to the distribution of mating success, as high-condition males secure most matings (Rowe & Houle 1996). For females, however, the narrative is different. Given that female fitness is gamete limited, not mate limited, the distribution of female reproductive success (fecundity) will tend towards normality (Fig. 2b). In other words, an overwhelming majority of females will reproduce, and variation in female condition will predict variation in fecundity, as opposed to variation in reproductive failure in the case of males. In sum, male fitness is more sensitive to variation in condition than female fitness (Fig. 3a).
It follows that a sex by environment interaction for fitness will always occur if the environment predicts condition. Most importantly, the interaction occurs in the simplest possible case, which is when the environment affects the condition of males and females in exactly the same way. The interaction occurs because males in high condition can generally expect to have greater fitness than females in the same condition, but females in low condition can generally expect to have greater fitness than males in the same condition. A parent should therefore adjust the sex of offspring if offspring condition is predictable: if high condition is predicted, then sex should be adjusted towards males, as a high-condition male is likely to provide greater fitness return than a high-condition female (Trivers & Willard 1973).
Trivers and Willard (1973) originally envisioned a polygamous mammalian system, perhaps because male-male competition for mates is obvious and intense in these species. Yet, these classical principles have since been extended to a variety of other taxa (West 2009). Indeed, an exponential distribution of male mating success – a key assumption of Trivers-Willard – is not restricted to mammals with extreme male–male combat or species with male-biased size dimorphism. The pattern is general (Arnqvist & Rowe 2005), and it is found in territorial reptiles (Trillmich 1983) as well as in other ectothermic vertebrates with no obvious signs of territorial behaviour or extreme male combat (Joneset al. 2002). I argue that these classic principles of sexual selection and condition-dependence also have an inexorable bearing on the evolution and maintenance of TSD. Specifically, if incubation temperature has, on average, an effect on the condition of individuals at adulthood, then the key requirement of the Charnov-Bull model, a sex by incubation temperature interaction for fitness, must occur when males are produced under favourable incubation temperatures and females under unfavourable incubation temperatures (Fig. 3b,c). The main difference between the hypothesis I propose and the Trivers-Willard hypothesis is that condition is predicted by temperature, instead of some aspect of the parental environment or phenotype (Fig. 3). This important difference also happens to be a strength of the hypothesis I propose, as it allows circumvention of a persistent criticism of Trivers-Willard: a well-understood proximate mechanism that can trigger conditional allocation to sex (Cameron et al. 2008). To differentiate my hypothesis from classic Trivers and Willard (1973), I refer to my hypothesis simply as the ‘Mighty Males’ hypothesis.
Trivers-Willard is the first general model of conditional sex allocation, so it is not surprising that Trivers-Willard features prominently in discussion of ESD (e.g., West 2009). Indeed, a few explanations for TSD in reptiles have made passing reference to, or are partly rooted in, the Trivers-Willard hypothesis (Head et al.1987; Deeming et al. 1988; Deeming & Ferguson 1989; Ewertet al. 1994; Roosenburg 1996). West (2009, p. 246) even goes so far as to state “One way this [sex-by-environment interaction] can occur is if one sex gains more than the other from developing in a “better” environment, as in the classic Trivers and Willard hypothesis”, but this statement is followed only by a brief discussion of studies that are limited to crocodilians, in which males guard harems as in classic Trivers-Willard (Deeming et al. 1988; Deeming & Ferguson 1989). What is perhaps surprising is that no study to my knowledge has explored the possibility that Mighty Males may provide a general explanation for TSD. I suggest the reason is because Trivers-Willard predates the Charnov-Bull model of ESD, it is not immediately obvious that the non-specific form of sex-by-environment interaction for fitness described by Charnov and Bull can be reconciled with sex-specific condition dependence for fitness, described by Trivers and Willard.