Introduction
Environmental sex determination (ESD) occurs in a phylogenetically diverse group of organisms, including plants, invertebrates, fish, and reptiles. Unlike genetic sex determination (GSD), where sex is determined by genetic factors at conception, sex under ESD is determined by an environmental cue experienced by the embryo after conception. It is perhaps easy to envision how frequency-dependent selection favours GSD when each sex contributes half their genes to the next generation, but the evolution of maintenance of ESD is puzzling given the possibility of extreme sex ratio bias that would favour the rarer sex (Bull 1980). The most promising and general explanation for ESD is known as the Charnov-Bull model, where ESD may be adaptive when there is a sex by environment interaction for fitness (Charnov & Bull 1977). In other words, male fitness and female fitness must respond differently to the environment, and ESD may evolve if a reliable cue can predict the environment during the embryonic period. Consistent with the Charnov-Bull model, photoperiod affects sex in the amphipodGammarus duebeni , because males benefit from increased size more than females, and hence males are born early in the season and grow to a bigger size (McCabe & Dunn 1997). In a parasitic wasp (Lariophagus sp) , eggs laid in large insect hosts become female, not male, as female wasps benefit more than males from increased size (fecundity) at adulthood (Charnov et al. 1981). Clearly, different environmental stimuli and various adaptive mechanisms can be consistent with the Charnov-Bull model. For specific types of ESD, then, the current challenge is to apply the Charnov-Bull model in a way that can generally explain the evolution and maintenance of ESD where it is taxonomically widespread.
The most common form of ESD is temperature-dependent sex determination, or TSD, which is found mainly in reptiles, but also in several fishes (Ewert et al. 1994; Janzen & Phillips 2006). In reptiles with TSD, the incubation temperatures experienced by the embryo during the thermosensitive period influences sex, where the thermosensitive period comprises specific anatomical stages that occur roughly during the middle third of embryonic development (Yntema 1968, 1979; see also Girondot et al. 2018). Under constant temperature, the temperature-sex reaction norm is known to take three forms. The FMF pattern occurs when males are produced at intermediate temperatures and females are produced at extreme temperatures (Fig 1a). The MF pattern occurs when males are produced at cool temperatures and females at hot temperatures (Fig 1b). Finally, the FM pattern occurs when females are produced at cool temperatures and males at hot temperatures (Fig. 1c). The FMF pattern is hypothesized to be ancestral, as it subsumes the FM and MF patterns.
Because TSD is phylogenetically widespread in reptiles, an adaptive explanation that applies to most or all afflicted species has been sought (reviewed in Janzen and Paukstis 1991; Shine 1999; Valenzuela 2004; Janzen and Phillips 2006), with the general presumption that any unifying explanation would be rooted in the Charnov-Bull model (Shine 1999). Despite considerable effort (Ferguson & Joanen 1983; Ewertet al. 1994; Janzen 1995; Morjan & Janzen 2003; Warner & Shine 2005; Spencer & Janzen 2014), there is no adaptive explanation for TSD that is both broadly convincing and supported (Shine 1999; Janzen & Krenz 2004; Schwanz et al. 2016). The root of the problem is likely related, in part, to the fact that temperature is not only a cue that affects sexual differentiation, but also has profound developmental and physiological effects that encompass a myriad of other traits expressed across an individual’s lifetime (Noble et al. 2018b). Uncovering the adaptive significance of TSD is therefore not as simple as in other ESD cases, such as photoperiodic cues, as temperature exhibits direct effects on many traits, including fitness.
Because an adaptive and broadly-applicable explanation for TSD has yet to be uncovered, the current paradigm is, arguably, that TSD persists in many reptiles because it is selectively neutral, and it is difficult to evolve GSD (Bull 1980; Girondot & Pieau 1999; Janzen & Krenz 2004). This hypothesis seems at odds with the recent evolution of GSD in several turtle lineages (Ewert & Nelson 1991), and with theory suggesting that transitions between GSD and TSD require little genetic innovation (Quinn et al. 2011). The hypothesis is also at odds with common and extreme biases in population sex ratios that would advantage the rarer sex (e.g., Ferguson and Joanen 1983; Schwarzkopf and Brooks 1985; Jensen et al. 2018), as well as rapid post-Wisconsinian evolution of TSD patterns with respect to climate (Ewert et al.2004, 2005; but see Carter et al. 2019a). In the present study, I point out that a simple and surprisingly well-supported explanation for a sex-by-environment interaction under TSD has been overlooked, one that predates the Charnov-Bull model. Ultimately, I integrate classical theories in evolutionary ecology to develop a broadly applicable framework that explains the evolution and maintenance of TSD. While I focus on patterns in reptiles, the explanation may also apply to some or all TSD fishes.