FREDRIC J. JANZEN AND JAMES G,KRENZ Phylogenetics Which was First, TSD or GSD? The basic challenge of evolutionary biology is to explain ever, that this temperature-dependent sex determination variation or the lack thereof, be it phenotypic, genetic, phy- (TSD) does not imply a lack of genetic involvement in sex logenetic, spatial, temporal, andso on. To illustrate, onegross determination, just an absence of sex chromosomes and generalization is that phenotypic traits we think of as being temperature-insensitive sex-determining genes (Valenzuela very important to organisms tend to be highly conserved et al. 2003). TSD can be contrasted with the more familiar (e.g., binocular vision in vertebrates), probably because genotypic sex determination (GSD), sometimes referred to the genomic and developmentalunderpinnings are essentially as chromosomal sex determination (Deeming and Fergu- fixed. Thus. one striking feature about sex-determining son 1988), wherein sex is fixed permanently by generic fac- mechanisms (SDMs), a fundamental aspect of sexual or- tors at conception. Exceptions exist (e.g., sex-changing ganisms, is the enormous variety (Bull 1983). fish), but these two categories serve to organize, identiiy, This great diversity of SDMs (in vertebrates in particu- and elucidate the key evolutionary issues involved (Valen- lar) has long puzzled biologists. Given the existence of sex- zuela et al. 2003). ual reproduction and its demonstrable adaptive significance In the course of cataloguingthis diversity, a controversy of (e.g., West et al. 1999), one might assume that the means by sorts has ensued over the origins of various sex-determining which it is accomplished might be highly conserved. More- systems in vertebrates. How many times has each mecha- over, SDMs and the primary sex ratio are inexnicably linked nism evolved and, as exemplified in the chapter title, which (Bd 1983), and selection for a 1:l primary sex ratio is mechanism is ancestral? The answers to these and related strong (Fisher 1930); thus, one might also assume that any questions strike to the heart of at least two critical and con- SDM producing a skewed primary sex ratio would be non- troversial issues involving SDMs: (1) adaptive significance existent. But these two assumptions are, remarkably, incor- and (2) molecular genetic and physiological underpinnings. rect. In fact, sex is determined in various vertebrate taxa Traits as fundamental as SDMs are typically not amenable in an extraordinary variety of ways (Bull 1983;Janzen and to experimental evolution, necessarily leaving informed PaMs199la). speculation and theoretical exploration to fill the void. In- SDMs in vertebrates nonetheless can be collected roughly deed, such has been the case regarding evolutionary transi- into two major categories. One is environmental sex deter- tions among SDMs in vertebrates (Ohno 1979; Bull 1983; mination (BSD), wherein sex is fixed permanently by envi- Karlin and kssard 1986; Ewert and Nelson 1991; Janzen ronmental cues (primarily temperature in vertebrates) during and Paukstis 1991a; Solari 1994; but see MUk and Green a discrete period after fertilization (Bull 1983). Note, how- 1990; Janzen and Paukstis l99lb; Kraak and Pen 2002). However, the development of rigorous, phylogeneticdy fish primarily have TSD Ib, crocodilians TSD 11, sphen- based comparative methods over the past two decades pro- odontians TSD Ib (or possibly TSD 11), lizards TSD Ib and vides another valuable approach: the retrodicrion of the TSD 11, and turtles TSD la and TSD II. evolutionary history of SDMs.This method is not perfect of course (e.g., it is correlative in nature), but it does pro- Hypotheses on Evolutionary Transitions vide a strong framework for much modern evolutionary re- search on the nature of variation (e-g., Harvey et al. 1996; Scenarios regarding evolutionary transitions between SDMs Martins 1996; Avise 2000; Page and Holmes 1998). in vertebrates have historically been qualitative. For es- In this chapter, the evolutionary history of SDMs in ver- ample, Witschi (1959) stated that fish had morphologically tebrates will be explored. First previous thinking on this indistinguishable (i-e., homomorphic) sex chromosomes, if topic will be discussed to illustrate the issues involved. Then any at all, representing "a primitive and ancient condition" comparative methods will be applied to rigorous (mainly (see also Ohno 1979), with genetically based sex determha- molecular) phylogeneuc hypotheses 111 vertebrates and par- uon thus arising in tetrapods in the Jurassic around 150 ticular vertebrate lineages, with special emphasis on lepi- million years ago. Similarly, Ohno (1967) proposed inde- dosaurs (lizards, snakes, andsphenodontians) and turtles, to pendent evolutionary transitions from ancestral homomor- evaluate evolutionary transitions between SDMs. These phic to derived heteromorphic sex chromosomes in lizards, comparative analyses will provide explicit tests of hypothe- snakes, and birds and a comparable more recent event in ses concerning the evolutionary diversity of SDMs in verte- mammals. These kinds of scenarios are understandable in brates. After the hypothesized ancestral SDM in vertebrates the absence of (1) information on the true diversity of is identified and the approximate phylogenetic loci of em- SDMs in vertebrates, (2) rigorous phylogenetic hypotheses, lurionary transitions between SDMs noted, the implica- and (3) acceptable formal comparative or molecular meth- tions of these results for important biological issues involv- ods. Furthermore, these propositions have no doubt per- ing TSD will be discussed. In particular, this discussion will sisted and become established dogma owing to support focus on the ramifications of these findings for research into from some population genetic models of the evolution of the adaptive significance and mechanistic underpinnings of sex chromosomes (e.g.,Chariesworth 1978; reviewed in Bull TSD in vertebrates. 1983; Charlesworth 1991) and a few indirect observationsof sex chromosome degradation after autosoma1 conversion Taxonomic Distribution (e.g., reviewed in White 1973). To the authors' knowledge, however, these verbal scenarios remain untested explicitly SDMs are diverse and nonrandomly distributed in verte- using modern comparative methods. brates. Few fish (see Conover, Chapter 2)and no amphibians The qualitative nature of hypotheses concerning the (Hayes 1998; Chardard et aL, Chapter 7), snakes (Janzen and evolution of SDMs in vertebrates has characterized most Paukstis 199la), birds, or mammals (Bull 1983) are known scenarios involving TSD as well. Most authors have weighed to have TSD. Instead, male and female heterogamety are in cautiously on. this subject, which is summed up nicely as, "widely distributed in fish and amphibians, whereas snakes "There is no dear empirical evidence suggesting the evolu- and birds only have female heterogamety, and mammals tionary order of [TSD] compared to GSD .. ." (Karlin and have only male heterogamety Crocodilians and sphenodon- Lessard 1986; see also Bull 1983;Janzen and Pauksris 19Pla). tians have exclusively TSD (Deeming, Chapter 4; Nelson et Kraak and Pen (2002) qualitatively interpreted their phylo- al., Chapter 6), but lizards and turtles exhibit a variety of genetic representation of SDMs in vertebrates to indicate genotypic and environmental SDMs (Bull 1983;Janzen and that both GSD and TSD have evolved multiple times (see Paukstis 1991a; Harlow, Chapter 5; Ewert et al., Chapter 3). also Janzen and Paukstis I991a) but staked no position re- GSD is much more common in lizards than in turtles; the garding their polarity. Even so, an underlying sentiment converse is true for TSD (Janzen and Paukstis 1991 a). More- regarding the ancestral nature of TSD (at least in reptiles) over, TSD is not monotypic; in fact, three types have been is evident (Karlin and Lessard 1986; Webb and Cooper- recognized, based on laboratory incubation of eggs (Bull Preston 1989; Janzen and Paukstis 199la; Solan 1994). On 1983; Conowr 1984; Ewert and Nelson 1991; Ewert et al. the other hand, some workers have cautioned against too 1994; Viets et d. 1994; Deeming, Chapter 4; Nelson et al., quickly rejecting an ancestd GSD scenario for reptiles Chapter 6; Conover, Chapter 2). The distribution of these (Bull 1983;Janzen and Paukstis 1991a). types of TSD, like that of SDMs in general, is nonrandom: To the authors' knowledge, only two studies have at- PHTLOGBNETICS: WHICH WAS FIRST, TSD OR GSDt 123 tempted quantitative tests of these verbal hypotheses, and several major nodes, particularly where sauropsids (reptiles at that only for turtles. Ewert and Nelson (1991), dung Gaff- and birds) split from mammals. Even with this poor resolu- ney's (1984) morphology-based turtle phylogeny, inferred tion at deep branches in the tree, several firm conclusions that TSD must have been lost independently four to six times can be drawn. First and foremost, GSD is an ancient con- in this group. Janzen and Paukstis (1991a), using phyloge- dition in vertebrates, and there is at least one dearly docu- netic information accumulated for another study (Janzen mented origin of TSD in fishes (Conover,Chapter 2). How- and Paukstis 199lb) and employing a parsimony-based ever, how early TSD arose 'within sauropsids cannot be comparative analysis, found that TSD was the ancestral determined unambiguously by this particular analysis. None- condition for turtles . The authors now explore these issues theless, TSW is nearly ubiquitous