186 0169-5347/99/$ – see front matter © 1999 Elsevier Science. All rights reserved. PII: S0169-5347(98)01575-4 TREE vol. 14, no. 5 May 1999

REVIEWS

Environmental sex determi-nation (ESD) occurs inboth plants and animals,and has stimulated consid-

erable research and speculationfrom evolutionary biologists1–3.Although some examples of ESDfit well with theory1,4–8, other casesremain an evolutionary enigmaand continue to be the focus of significant controversy3,6,9–12.Temperature-dependent sex deter-mination (TSD) in reptiles pro-vides such a case. Several alterna-tive explanations for the nature ofthe adaptive advantage conferredby TSD have been proposed4,13,14.All of these hypotheses invoke aseries of assumptions, most ofwhich had little empirical support at the time the hypoth-eses were generated. This situation has now changed con-siderably, with a spate of recent papers describing experi-mental studies that clarify a fundamental underpinning ofthe theoretical models – reaction norms of reptilian embryosin response to the physical conditions under which theyare incubated. Hence, we can now look more closely atcompeting hypotheses for the adaptive significance of TSDin reptiles. My aim is to identify the unique features of eachof these hypotheses and evaluate the assumptions uponwhich they are based in the light of recent experimentalstudies. In particular, how do incubation temperaturesinfluence organismal fitness in these animals?

Hypotheses for the adaptive significance of TSDAlthough many authors have speculated on the evolu-

tionary advantages of TSD in reptiles, the hypotheses fallinto a few clear categories:• Phylogenetic inertia: TSD is an ancestral form of sexdetermination with no current adaptive significance. Thishypothesis is difficult to test but seems inconsistent with thephylogenetic lability of sex-determining systems in reptiles13.• Group-adaptation: TSD has evolved because it facilitatesgroup fitness through sex-ratio skewing13. This hypothesisrelies upon the onerous assumptions inherent in groupselection15, is an argument for sex-ratio bias rather thanany specific sex-determining system and is not consistentwith available data on population structure of TSD reptiles14.• Inbreeding avoidance: by resulting in single-sex clutchesfrom natural nests, TSD reduces deleterious effects ofinbreeding among siblings13. This hypothesis is inconsist-ent with the prevalence of TSD in long-lived iteroparousorganisms (such as crocodilians and many turtles), whereoffspring from many annual cohorts interbreed14.• Differential fitness: if incubation temperatures differen-tially affect the fitness of male and female offspring, TSDcan enhance maternal fitness by enabling the embryo todevelop as the sex best-suited to those incubation condi-tions1. This hypothesis is logically more robust than the

preceding ones and seems themost consistent with availabledata3,9,16. However, it has notbeen widely appreciated that the‘differential fitness’ models actu-ally include a considerable diver-sity of separate hypotheses thatadvocate quite different mecha-nisms as to why sex and incu-bation temperatures can have aninteractive effect on organismalfitness. Table 1 presents a classifi-cation of these models and anattempt to characterize assump-tions underlying each of them.

The six models listed in Table 1differ in significant ways. For ex-ample, the phenotypic variancethat is acted upon by selection is

generated by fundamentally different mechanisms: bymaternal effects (egg size) in the ‘size-matching’ model(Table 1a); by seasonal shifts in hatching date in the ‘time-matching’ model (Table 1b); by incubation temperatures(with the sexes showing identical reaction norms) in the‘nest-philopatry’ and ‘phenotype-matching’ models (Table1c,d); and by the interaction between sex and incubationtemperature (with the sexes following different norms ofreaction) in the ‘interaction’ models (Table 1e,f). The puta-tive mechanisms by which organismal fitness is affectedalso vary, and might involve differential mortality (either inthe egg or in later life-history stages) or phenotypic modifi-cations that translate into viability differences (again, at arange of life-history stages) between males and femalesfrom different incubation conditions.

All of the models posit links between incubation tem-perature and phenotype, and between phenotype and fit-ness. All rely on males and females differing in the overallrelationship between incubation temperature and fitness,but this sex difference is suggested to emerge in differentways. For some, the link that differs between the sexes is thatbetween phenotype and fitness (Table 1a–d), whereas oth-ers rely on sex differences in the effects of incubation tem-perature on phenotype (the interaction models, Table 1e,f).

Validity of the assumptionsHow realistic are the assumed mechanisms linking

incubation temperature to male versus female fitness inthese models? This empirical question can be asked as eas-ily of a species with genetic sex determination (GSD) asone with TSD. Indeed, it is notable that the adaptationisthypotheses outlined in Table 1 bear strong similarities inlogical structure to analogous hypotheses for sex-ratiomanipulation in GSD species. In many cases, the hypoth-eses based on GSD sex-ratio manipulation have substantialempirical support. For example, the size-matching model(Table 1a, sex-ratio adjustment relative to offspring size)bears a strong similarity to explanations for the ability offemale invertebrates (e.g. hymenopterans17) to adjust

Why is sex determined by nesttemperature in many reptiles?

Richard Shine

Temperature-dependent sex determination(TSD) is widespread in reptiles but its

adaptive significance remainscontroversial. The most plausible

theoretical models for the evolution of TSDrely upon differential fitness of male and

female offspring incubated under differentthermal regimes. Several mechanisms

might generate these fitness differentials,and recent work provides empirical

support for some of them.

Richard Shine is at the School of Biological Sciences,Building A08, The University of Sydney, NSW 2006,

Australia (rics@bio.usyd.edu.au).

TREE vol. 14, no. 5 May 1999 187

REVIEWS

offspring gender based on the probable size of their off-spring at birth. The time-matching model (Table 1b, sex-ratio adjustment relative to date of hatching) resemblesadaptive explanations for the correlation between hatch-ing date and offspring sex ratio in GSD organisms asdiverse as birds18 and fig wasps2. The nest-philopatrymodel (Table 1c, sex-ratio adjustment in relation to sex dif-ferences in the degree of philopatry and consequently inthe probability of ‘inheritance’ of the nest site) is analo-gous to sex-ratio manipulations in some primate societies,whereby the nondispersing sex (daughter) is overpro-duced by high-status females that can transfer their statusto their nondispersing offspring19. The phenotype-matchingmodel (Table 1d, overproduction of the sex that derives the greatest benefit from the available phenotypicconditions at hatching) is similar to a situation in verte-brates in which offspring sex ratio is manipulated relativeto maternal (and thus offspring) condition19,20. It is moredifficult to find parallels with the ‘interaction’ models(Table 1e,f), because they invoke a direct interactionbetween sex and incubation temperature. Importantly, the models listed in Table 1 do not provide mutually exclu-sive explanations, and it is theoretically possible for all six of these models to apply simultaneously to the samepopulation.

I focus on differential-fitness (Charnov–Bull) modelsrather than alternative approaches (such as group adap-tation and sib-avoidance), for reasons already mentioned. In the following sections, I consider the models listedin Table 1 and discuss the empirical support available for assumptions that are invoked by each of them. Anyadaptationist hypothesis is based upon a massive array of assumptions about the natural world and the efficacy ofnatural selection, and consequently I can deal with only a small subset of them. Hence, I will highlight assumptionsconcerning the ways in which incubation temperaturesmight influence organismal fitness in reptiles, rather than,

for example, the occurrence of sex-specific philopatry(nest-philopatry model, Table 1c). The following assump-tions are of particular interest.

Phenotype at hatching affects organismal fitnessA link between hatchling traits and lifetime reproduc-

tive success (LRS) is assumed by many adaptationist mod-els, including all those listed in Table 1. (Model e, the sex-by-temperature-interaction’ model, which relies uponmortality prior to hatching, offers a possible exceptionbecause differential incubation-dependent fitness isalready expressed at hatching and need not occur later inthe life history. However, the issue is a matter of definition,and whether one is prepared to accept ‘hatches versusdoes not hatch’ as a phenotypic trait of the embryo.) Several studies have documented phenotypic correlates of survival rates in hatchling reptiles under field condi-tions10,21–24 but the generality of such effects remainsunknown. Although the models listed in Table 1 generallyfocus on phenotypic traits that are potentially measurableat hatching, the models do not require that all phenotypictraits affecting LRS are expressed early in ontogeny. Traits that were masked until maturation (e.g. the degreeof elaboration of secondary sexual characteristics) wouldsuit the models just as easily as traits such as offspring size or shape25.

One phenotypic trait of particular interest is time athatching – the central feature of the time-matching model(Table 1b). Field studies on many kinds of organism haverevealed strong fitness advantages associated with hatch-ing early in the season26–28. The most convincing empiricalsupport for adaptive advantages to TSD involves time:early-hatching fish (Menidia menidia) grow larger beforebreeding, and the consequent size-induced increase in LRSas a result of early hatching can be higher for females thanfor males7,8. Under these conditions, the observed produc-tion of females early in the season (in response to water

Table 1. An overview of ‘differential fitness’ models for the adaptive significance of temperature-dependent sex determination (TSD) in reptilesa

Source of phenotypic Why does incubation temperatureRole of incubation temperature variation that engenders affect fitness differently in

Model (apart from sex determination) fitness variation males and females? Refs

(a) Matching sex to egg size None: simply enables female to Maternal effects (egg size) Egg size affects fitness differently 12determine sex ratio of her offspring in males and females, and TSD

enables mother to adjust sex ratiorelative to offspring size.

(b) Matching sex to time of hatching Correlate (predictor) of seasonal Seasonal timing of hatching Hatching time affects fitness 8timing of hatching differently in males and females,

and TSD enables mother to matchoffspring sex ratio to time of hatching.

(c) Nest philopatry Influences offspring fitness Incubation temperature Because of sex differences in philopatry, 44(e.g. survival, growth and viability) daughters derive more advantage fromindependent of sex a ‘good’ nest (which they eventually use

themselves) than do sons.(d) Matching sex to phenotype Induces changes to phenotype, Incubation temperature Incubation temperature affects 3

independent of sex phenotype, and phenotypic determinantsof fitness differ between males and females.

(e) Sex by temperature interaction Influences survival of eggs Interaction between Differential mortality of males and 42for offspring survival or hatchlings, but differently incubation temperature females from different incubation

in males and females and sex temperatures.(f) Sex by temperature interaction Induces changes to phenotype, Interaction between Thermally induced phenotypes affect 16,25for offspring phenotypes but differently in males and females incubation temperature fitness, with the relationship between

and sex phenotype and fitness identical in both sexes.

aNote that references are examples only and do not necessarily refer to the first mention of the model in published literature.

188 TREE vol. 14, no. 5 May 1999

temperatures at that time) might offer a fitness advantageto reproducing females that exhibit TSD rather than GSD(Refs 7,8). Similar fitness advantages from early hatchingare plausible for many reptilian systems29. It will often betrue that hotter nests hatch earlier than cooler nests, atleast in those species with a single clutch per year, andthat ‘early’ and ‘late’ nests will differ thermally in specieswith multiple clutches. Hence, covariation between hatch-ing date and nest temperature is likely to be common infield populations of reptiles.

Phenotype at hatching affects organismal fitnessdifferently in males and females

The first four models listed in Table 1 make thisassumption. Field studies provide abundant evidence thatphenotypic traits contribute differently to LRS in malesand females, and hence that selection pressures on pheno-typic traits can differ substantially between conspecificmales and females30. However, I do not know of any empiri-cal data on reptiles to link hatchling phenotypes to event-ual LRS. Also, theoretical models that posit an adaptive sig-nificance of TSD require that this relationship differs notonly between the sexes, but also in such a way that TSDwill maximize offspring fitness. It remains possible thatincubation conditions affect phenotypes and that pheno-types affect LRS – but that the direction(s) of those influ-ences do not result in maternal fitness being maximized bythe particular pattern of sex ratio versus temperature generated by TSD.

Incubation temperature affects the phenotype athatching

This assumption is clearly part of three models in Table 1(c, nest-philopatry; d, phenotype-matching; and f, the sex-by-temperature-interaction model that deals with pheno-typic traits). An additional model (the sex-by-temperature-interaction model that relies upon differential mortalityprior to hatching, Table 1e) also fits into this categorybecause the differential mortality is presumably a result ofsome unmeasured component of the phenotype. Recentwork supports the generality of incubation effects onhatchling phenotypes in many reptilian taxa, includinglizards, snakes, crocodilians and turtles. These studiesinclude work on both GSD species31–35 and TSD species36–38.Most of the data for TSD species can be challenged on thegrounds that apparent effects of incubation temperaturesare (or might be) a secondary consequence of sex. How-ever, recent work has used hormonal manipulation tooverride thermal cues for sex determination and clearlydemonstrate direct effects of incubation temperatures ongrowth rates of juvenile turtles in a TSD species3.

Nonetheless, many questions have yet to be resolved.For most of these systems, we do not know if biologicallymeaningful levels of phenotypic variation are generated bythe range of nest temperatures that the organisms experi-ence under natural conditions39–41. Natural nests might notvary enough to induce long-lasting phenotypic effects ontraits that actually influence LRS. Testing this assumptionremains a major challenge for field biologists.

Thermally induced differences in survival rates of eggsin natural nests might also constitute an important sourceof temperature-induced differentials in organismal fitness.However, the magnitude of variance in egg survival intro-duced by temperature could be minor compared with thevariance introduced by other factors (e.g. soil moistureand predation). Further data on the determinants of hatch-ing success in natural nests are needed.

Male and female phenotypes at hatching are affecteddifferently by incubation temperature

This assumption is inherent in only two of the modelslisted in Table 1: the sex-by-temperature-interaction mod-els (e and f). Somewhat ironically, the best data in supportof this assumption come from studies on GSD snakes andlizards. Burger and Zappalorti42 demonstrated sex-specificmortality of snake eggs at different incubation tempera-tures. The phenotypic responses of hatchling scincid lizards(Bassiana duperreyi) to incubation temperature indicatethat males and females have different thermal optima for incubation16. In one TSD species of gekkonid lizard(Eublepharis macularius), growth rates are affected by bothsex and incubation temperature25.

ProspectsOverall, there are several plausible pathways by which

differential-fitness models might apply to reptilian popu-lations and favor the evolution of TSD. Although manyassumptions of these models remain untested, the generalimpression from recent work is that incubation tempera-tures can substantially affect organismal traits that arelikely to be related to LRS; and that the nature of this effectcould differ between the sexes. These results are encour-aging for those who advocate an adaptive significance forTSD in reptiles.

Stronger tests among competing models will requiredata sets on the nature of the link between incubation envi-ronment and LRS. Two problems that precluded such workhave recently been solved. First is the discovery of increas-ing numbers of model systems that are more amenable toexperimentation than those that have been used tradition-ally. Initial research on the adaptive significance of TSDfocused primarily on crocodilians and turtles. The delayedmaturation and long reproductive lifespan of these animals(with other logistical constraints) effectively precludedexperimental studies of the ways that incubation condi-tions might influence organismal fitness. The discovery ofTSD in shorter-lived organisms6,25,43 overcomes this ob-stacle. The second problem is the confounding effect of sexand incubation temperature (i.e. some temperatures pro-duce males, others produce females), making it difficult toseparate temperature effects from gender effects. Hor-monal manipulation can break this link3,25 and thus providean opportunity for direct experimentation to test the cen-tral tenet of adaptationist models in this field: the notionthat a female who can adjust her offspring’s sex to theirincubation temperature can thereby enhance her owngenetical fitness.

AcknowledgementsI thank P. Harlow and M. Elphick for discussions and theAustralian Research Council for financial support.

References1 Charnov, E.L. and Bull, J.J. (1977) When is sex environmentally

determined? Nature 266, 828–8302 Charnov, E.L. (1982) The Theory of Sex Allocation, Princeton

University Press3 Rhen, T. and Lang, J.W. (1995) Phenotypic plasticity for growth

in the common snapping turtle: effects of incubation temperature, clutch, and their interaction, Am. Nat. 146, 726–747

4 Bull, J.J. (1980) Sex determination in reptiles, Q. Rev. Biol. 55, 4–215 Bull, J.J. and Charnov, E.L. (1988) How fundamental are Fisherian

sex-ratios? in Oxford Surveys in Evolutionary Biology (No. 5) (Harvey, P.H. and Partridge, L., eds), pp. 96–135, Oxford University Press

REVIEWS

TREE vol. 14, no. 5 May 1999 189

REVIEWS

6 Bull, J.J., Gutzke, W.H.N. and Bulmer, M.G. (1988) Nest choice in acaptive lizard with temperature-dependent sex determination,J. Evol. Biol. 2, 177–184

7 Conover, D.O. and Kynard, B.E. (1981) Environmental sexdetermination: interaction of temperature and genotype in a fish, Science 213, 577–579

8 Conover, D.O. (1984) Adaptive significance of temperature-dependent sex determination in a fish,Am. Nat. 123, 297–313

9 Janzen, F.J. and Paukstis, G.L. (1991) Environmental sexdetermination in reptiles: ecology, evolution and experimentaldesign, Q. Rev. Biol. 66, 149–179

10 Janzen, F.J. (1993) An experimental analysis of natural selection on body size of hatchling turtles, Ecology 74, 332–341

11 Janzen, F.J. (1996) Is temperature-dependent sex determination inreptiles adaptive? Trends Ecol. Evol. 11, 253

12 Roosenburg, W.M. (1996) Maternal condition and nest site choice:an alternative for the maintenance of environmental sexdetermination? Am. Zool. 36, 157–168

13 Ewert, M.A. and Nelson, C.E. (1991) Sex determination in turtles:diverse patterns and some possible adaptive values,Copeia 1991, 50–69

14 Burke, R.L. (1993) Adaptive value of sex determination mode andhatchling sex-ratio bias in reptiles, Copeia 1993, 854–859

15 Williams, G.C. (1966) Adaptation and Natural Selection: a Critique of Some Current Evolutionary Thought, Princeton University Press

16 Shine, R., Elphick, M.J. and Harlow, P.S. (1995) Sisters like it hot,Nature 378, 451–452

17 Werren, J.H. (1983) Sex-ratio evolution under local matecompetition in a parasitic wasp, Evolution 37, 116–124

18 Dijkstra, C., Daan, S. and Buker, J.B. (1990) Adaptive seasonal variation in the sex-ratio of kestrel broods, Funct. Ecol. 4, 143–147

19 Clutton-Brock, T.H. and Iason, G.R. (1986) Sex-ratio variation inmammals, Q. Rev. Biol. 61, 339–374

20 Trivers, R.L. and Willard, D.E. (1973) Natural selection of parentalability to vary the sex-ratio of offspring, Science 179, 90–92

21 Janzen, F.J. (1995) Experimental evidence for the evolutionarysignificance of temperature-dependent sex determination,Evolution 49, 864–873

22 Olsson, M., Gullberg, A. and Tegelstrom, H. (1996) Malformedoffspring, sibling matings, and selection against inbreeding in the sand lizard (Lacerta agilis), J. Evol. Biol. 9, 229–242

23 Madsen, T. and Shine, R. (1998) Quantity or quality? Determinants of maternal reproductive success in tropical pythons (Liasis fuscus), Proc. R. Soc. London Ser. B 265, 1521–1525

24 Jayne, B.C. and Bennett, A.F. (1990) Selection on locomotorperformance capacity in a natural population of garter snakes,Evolution 44, 1204–1229

25 Tousignant, A. and Crews, D. (1995) Incubation temperature andgonadal sex affect growth and physiology in the leopard gecko(Eublepharis macularius), a lizard with temperature-dependentsex determination, J. Morphol. 224, 159–170

26 Sinervo, B. and Doughty, P. (1996) Interactive effects of offspringsize and timing of reproduction on offspring reproduction:experimental, maternal, and quantitative genetic aspects,Evolution 50, 1314–1327

27 Spear, L. and Nur, N. (1994) Brood size, hatching order andhatching date: effects on four life-history stages from hatching to recruitment in western gulls, J. Anim. Ecol. 63, 283–289

28 Svensson, E. (1997) Natural selection on avian breeding time:causality, fecundity-dependent, and fecundity-independentselection, Evolution 51, 1276–1283

29 Olsson, M.M. et al. (1996) Paternal genotype influences incubationperiod, offspring size, and offspring shape in an oviparous reptile,Evolution 50, 1328–1333

30 Andersson, M. (1994) Sexual Selection, Princeton University Press31 Burger, J. (1998) Antipredator behaviour of hatchling snakes:

effects of incubation temperature and simulated predators,Anim. Behav. 56, 547–553

32 Phillips, J.A. and Packard, G.C. (1994) Influence of temperature and moisture on eggs and embryos of the white-throated savannamonitor Varanus albigularis: implications for conservation,Biol. Conserv. 69, 131–136

33 Reichling, S.B. and Gutzke, W.H.N. (1996) Phenotypic consequencesof incubation environment in the African elapid genusAspidelaps, Zoo Biol. 15, 301–308

34 Shine, R. and Harlow, P.S. (1996) Maternal manipulation of offspringphenotypes via nest-site selection in an oviparous lizard,Ecology 77, 1808–1817

35 Van Damme, R. et al. (1992) Incubation temperature differentiallyaffects hatching time, egg survival and sprint speed in the lizardPodarcis muralis, Herpetologica 48, 220–228

36 Congdon, J.D., Fischer, R.U. and Gatten, R.E. (1995) Effects ofincubation conditions on characteristics of hatchling Americanalligators, Herpetologica 51, 497–504

37 O’Steen, S. (1998) Embryonic temperature influences juveniletemperature choice and growth rate in snapping turtles Chelydraserpentina, J. Exp. Biol. 201, 439–449

38 Flores, D., Tousignant, A. and Crews, D. (1994) Incubationtemperature affects the behavior of adult leopard geckos(Eublepharis macularius), Physiol. Behav. 55, 1067–1072

39 Wilson, D. (1998) Nest-site selection: microhabitat variation and its effects on the survival of turtle embryos, Ecology 79,1884–1892

40 Packard, G.C., Miller, K. and Packard, M.J. (1993) Environmentallyinduced variation in body size of turtles hatching in natural nests,Oecologia 93, 445–448

41 Shine, R., Elphick, M.J. and Harlow, P.S. (1997) The influence ofnatural incubation environments on the phenotypic traits ofhatchling lizards, Ecology 78, 2559–2568

42 Burger, J. and Zappalorti, R.T. (1988) Effects of incubationtemperature on sex-ratios in pine snakes: differential vulnerability of males and females, Am. Nat. 132, 492–505

43 Harlow, P. and Shine, R. Temperature-dependent sexdetermination in the frillneck lizard, Chlamydosaurus kingii(Agamidae), Herpetologica (in press)

44 Reinhold, K. (1998) Nest-site philopatry and selection forenvironmental sex determination, Evol. Ecol. 12, 245–250

• Animal evolution: the end of the intermediate taxa? Andr� Adoutte et al. Trends in Genetics 15, 104–108 (March 1999)

• How many fungi does it take to change a plant commu-nity? Tim J. Danielle et al. Trends in Plant Science 4, 81–82 (March 1999)

• Shoot apical meristems and floral patterning: an evolu-tionary perspective, Victor A. Albert Trends in Plant Science 4, 84–86 (March 1999)

• Exploring a novel perspective on pathogenic relation-ships, Jeffrey D. Cirillo Trends in Microbiology 7, 96–98(March 1999)

• Selection of drug-resistant HIV, P. Richard Harrigan and Christopher S. Alexander Trends in Microbiology7, 120–123 (March 1999)

• The ecological role of bacteriocins in bacterial competi-tion, Margaret A. Riley and David M. Gordon Trends in Microbiology 7, 129–133 (March 1999)

Current trendsÐ articles of ecological or evolutionary interest in recent issues of other Trends magazines