Oviparity or egg-laying is the dominant mode ofreproduction among vertebrates Nevertheless

viviparity the retention of the egg within the reproductivetract until embryonic development is completecharacterizes almost all mammals it has also had at least150 independent origins within the fishes amphibiansand reptiles (Shine 1985 Blackburn 1992 Wourms andLombardi 1992) These multiple origins suggest pervasivebenefits to viviparity across a wide range of taxa lifehistories and habitats In the squamate reptiles (lizardsand snakes) for example viviparity is the most commonreproductive mode in cold climates and recent origins ofviviparity in this group are also associated with coldclimates (Shine 1985) Gravid females in cold climates canthermoregulate to keep embryos warmer than they wouldbe in a nest thus enhancing development Thermo-regulation by the female may thus ensure that birth occursat the appropriate season or even that reproduction issuccessful at all Viviparity is also advantageous in very wetor dry habitats for example because it obviates the needfor females to find suitable sites in which to lay their eggs

In reptiles viviparity is associated with a plethora ofintegrated morphological and physiological features thatare not present in oviparous reptiles these features arepresumed necessary for successful embryonic develop-ment in the oviduct (Packard et al 1977 Guillette 1993)Early insights into the evolution of these reproductive fea-tures were based on comparisons between typicaloviparous and viviparous species Some of the distin-guishing features of the viviparous species examined werethe major reduction or absence of an eggshell and thepresence of some form of placentation (Weekes 1935 andincluded references) However because the species used inthese comparisons represented the extremes of a putativeevolutionary sequence their use as a model for elucidatingthe actual sequence or timing of the morphological andphysiological changes attending the evolution of vivipari-ty is limited In fact these observations are consistent withboth a saltation model that posits that the characteristicfeatures of viviparity arise suddenly and simultaneouslyand a gradualist model that posits incremental evolutionfrom one reproductive mode to the other (Blackburn1992 1995)

A relatively new and more powerful approach is clarify-

ing the evolution of viviparity this approach involves theidentification and critical evaluation of closely related taxathat vary in reproductive mode (Guillette 1982 Heulin1990 Mathies and Andrews 1995 Qualls 1996 Smith andShine 1997 Meacutendez-de la Cruz et al 1998) Of these stud-ies Qualls (1996) provides the best evidence that a grad-ual process describes the evolution of viviparity His studyon Lerista bougainvillii an Australian skink involved threeconspecific populations one oviparous another vivipa-rous and a third that is morphologically and physiologi-cally intermediate An increase in the length of egg reten-tion associated with reduction in eggshell thicknessamong these populations supports the hypothesis thatviviparity evolves gradually from oviparity The otherstudies presented data that are consistent with this inter-pretationmdashthat is conspecific populations or closelyrelated species exhibited the expected grade of featuresintermediate between oviparity and viviparity

In this article we present a broad and unconventionalperspective on the evolution of viviparity Our goal asphysiological ecologists is to determine ldquohow animals aredesigned with reference to their natural environments andevolutionary historiesrdquo (Bennett 1987) We thus suggestthat the key to understanding how viviparity evolves is tofirst understand why it has not evolved in particular taxa(see also Shine 1985 Packard et al 1989) Our perspective

March 2000 Vol 50 No 3 BioScience 227

Articles

Robin M Andrews (e-mail randrewsvtedu) is a professor in the

Department of Biology Virginia Polytechnic Institute and State

University Blacksburg VA 24061-0406 Tom Mathies is a postdoc-

toral researcher at the National Wildlife Research Center Fort

Collins CO 80521-2154 copy 2000 American Institute of Biological

Sciences

Natural History of ReptilianDevelopment Constraints on theEvolution of ViviparityROBIN M ANDREWS AND TOM MATHIES

A POWERFUL APPROACH IS CLARIFYING THE

EVOLUTION OF VIVIPARITY THIS APPROACH

INVOLVES THE IDENTIFICATION AND

CRITICAL EVALUATION OF CLOSELY RELATED

TAXA THAT VARY IN REPRODUCTIVE MODE

focuses on the consequences of extended egg retention onembryonic development in oviparous species and thusidentifies potential constraints on the shift from oviparityto viviparity (more precisely lecithotrophy sensu Black-burn 1992) This approach lends itself to understandingthe evolution of viviparity at high taxonomic levels (egturtles versus squamates) as well as at low taxonomic lev-els (within genera and species)

We begin by reviewing the general pattern of embryon-ic development of reptiles to set the framework for further

discussion We then summarize the timing of ovipositionin reptiles and how timing affects embryonic developmentin the oviducts If as we assume the transition betweenoviparity and viviparity involves a progressive increase inthe amount of development that takes place in theoviduct then any factor that limits development in theoviduct also limits the potential for viviparity to evolveWe therefore discuss a number of reasons why ovipositionoccurs at particular stages of embryonic development Wenext consider the consequences of extended egg retentionin a clade of phrynosomatid lizards and how these conse-quences may restrain or enhance the evolution of vivipar-ity in this group Finally we summarize evidence thathypoxia (oxygen deprivation) is the physiological factorthat limits development of embryos in the oviducts

Comparative embryologydevelopmental chronologyEmbryonic development is a continuum from fertilizationto hatching or birth For comparative purposes howeverembryologists break development into a series of discretestages that are demarcated by the appearance of distinctfeatures such as neurulation torsion pharyngeal cleftslimb buds and so on Ideally a ldquonormal tablerdquo of develop-ment would list these stages extending from fertilizationto hatching or birth and would report the timing of eventsat some specific temperature Such an ideal normal tableis however not available for any reptile For example themost commonly used normal table for squamates starts atcleavage and has 40 stages but these stages are not relatedto time (Dufaure and Hubert 1961) In contrast whereasnormal tables for crocodilians provide the timing of devel-opmental events for particular incubation temperaturesstaging does not start until oviposition (neurulation) andthe normal table has only 28 stages (Ferguson 1985)

At a general level however patterns of development arecomparable among reptilian taxa First differentiation israpid initially but slows appreciably as the embryoapproaches the end of development with embryos passingthrough 80 or more of their recognized stages byapproximately halfway through the developmental periodThe relationship between developmental stage and timefor crocodilian and sea turtle embryos illustrates this phe-nomenon (Figure 1) The finding that most stages arecompleted before mid-development irrespective of taxonreflects the fact that early development is characterized bytissue differentiation and organogenesis processes thatprovide many distinct landmarks and thus result in theidentification of many distinct stages Later developmentis largely characterized by growth in size and by biochem-ical and physiological changes that provide few new fea-tures for visual staging Indeed by mid-developmentembryos are small but recognizable lizards snakes croco-diles or turtles

A second general pattern of reptilian development isthat growth in size is exponential (Figure 2) The magni-

228 BioScience March 2000 Vol 50 No 3

Articles

Figure 1 The relationship between embryonic stage and timesince oviposition with time expressed as percentage of theincubation period in crocodilian and sea turtle embryos(top) For Crocodylus porosus (Ferguson 1985) staging startsat oviposition when the embryo is at the neurula stage(roughly equivalent to stage 14 for marine turtles) (bottom)For marine turtles (Miller 1985) staging starts at ovipositionwhen the embryo is at the gastrula stage The time betweenovulation and oviposition is estimated (Miller 1985) to be 8ndash9daysmdashequivalent to 14 of the time between oviposition andhatching (60 days) The shapes of the curves for both reptilesare similar rapid development during the first half ofincubation reflects tissue and organ differentiation andslower development later in development reflects an increasein embryo size and biochemical and physiological changesthat provide few new features for visual staging

tude of growth toward the end of development is consid-erably greater than depicted in Figure 2 for two reasonssize is indexed by length rather than mass and the x-axisrepresents stage rather than time That is if the axes weremass and time the curve would exhibit a much greaterincrease in size after mid-development than shown in Fig-ure 2 The second half of development is thus character-ized by a size increase of considerable magnitude

Stage at ovipositionFor oviparous reptiles oviposition marks the transitionbetween embryonic development within the mother and inthe environment The timing of this transition varies con-siderably among reptilian groups Eggs of turtles crocodil-ians and sphenodontidans are invariably laid whenembryos are in the earliest stages of development In con-trast the eggs of most squamates are laid when embryos areapproximately one-third through development althoughoviposition can occur substantially earlier or later As wediscuss several morphological and physiological factorsmay constrain the stage at oviposition and thus account forvariation in the reproductive modes of reptiles

Turtles crocodilians and sphenodontidans Ovi-position by turtles (Chelonia) and tuataras (Sphenodonti-da) occurs when the embryo is at the gastrula stage hencelittle development occurs in the oviduct (Ewert 1985 Mof-fat 1985) After reaching the gastrula stage turtle embryosenter developmental arrest resuming development onlyafter oviposition The amount of time that turtle embryosspend in developmental arrest varies Turtle eggs remainin the oviduct for at least the 2-week period that is neces-sary for shell formation However the length of egg reten-tion can vary considerably among females The longestperiods of egg retention for the North American sliderturtle Trachemys scripta and the mud turtle Kinosternonsubrubrum in South Carolina were 39 days and 50 daysrespectively (Buhlmann et al 1995) Some females of thechicken turtle Deirochelys reticularia retain their eggsover winter and lay them the following spring presumablybecause conditions in the fall were not favorable for suc-cessful nesting These eggs are thus retained by the femalefor 4ndash7 months before oviposition (Buhlmann et al 1995)and hatch in the spring of the following year (Congdon etal 1983) In contrast to all other reptiles a long period ofegg retention is characteristic of tuataras eggs are in theoviduct for 6ndash8 months before oviposition (Cree et al1992) Development is presumably arrested while eggs arein the oviduct because embryos are at only the gastrulastage at oviposition

Crocodilians oviposit when embryos are at the neurulastage and thus somewhat more advanced than those ofturtles at the time of oviposition (Ferguson 1985) Unliketurtle embryos however crocodilian embryos do notundergo developmental arrest and therefore must be laidimmediately Nevertheless oviposition at early develop-

mental stages appears to be characteristic of all turtlescrocodilians and sphenodontidans The fact that embryosof these taxa do not develop beyond an early developmen-tal stage while in the oviduct has apparently limited thereproductive options open to these taxamdashno turtle croc-odilian or sphenodontidan is viviparous

For turtles crocodilians and sphenodontidans theapparent requirement for oviposition to occur early indevelopment may reflect a suite of morphological andphysiological constraints that were established early in thehistory of these taxa The most immediate constraint onembryonic development may be the limited exchange ofgases particularly oxygen in the oviduct Gas diffusion isconsiderably slower in water than in air and the pores andchannels of the eggshell are filled with fluid when the eggis in the oviduct Moreover the heavily calcified shells ofthe eggs of crocodilians and most turtles (Packard andDeMarco 1991) may further exacerbate the problem of gasexchange Indirect evidence to support this hypothesis forturtles includes the observations that the shell calcifies justbefore gastrulation (when development becomes arrested)and that after oviposition embryos of species with morepermeable flexible-shelled eggs develop faster thanembryos of species with rigid calcareous shells (Ewert1985)

Experimental studies on the northern snake-necked

March 2000 Vol 50 No 3 BioScience 229

Articles

Figure 2 The relationship between embryo size and stagefor Lacerta vivipara a viviparous lizard Embryo size isindexed by its length which actually underestimates theactual size of the embryo (see text for details) Embryo stagefollows Dufaure and Hubert (1961) Stages at whichimportant developmental events occur (gastrulationneurulation appearance of limb buds and digitscompletely differentiated) the range of stages over whichoviposition occurs in the majority of oviparous squamatesand the stage of parturition in viviparous squamates areindicated Figure modified from Xavier and Gavaud(1986)

turtle Chelodina rugosa in Australia provide direct evi-dence that hypoxia maintains developmental arrest in thisspecies (Kennett et al 1993) The rigid-shelled eggs arelaid underwater in seasonal ponds and the embryosremain in developmental arrest as long as the eggs areimmersed in water Embryos resume development onlyafter the soil dries and oxygen tension rises In an experi-ment in which eggs were shifted from water to an atmos-phere of pure nitrogen embryos remained in develop-mental arrest they resumed development only whenexposed to atmospheric air

An alternative or perhaps complementary explanationfor oviposition early in development by turtles and croco-dilians concerns the adhesion of the embryo and thevitelline membrane to the shell (Ewert 1985 Ferguson1985) Adhesion shortly after oviposition appears to havea respiratory function because the ldquochalkingrdquo (drying) ofthe shell that is associated with adhesion increases its con-ductance to gases (Thompson 1985) Because movementof an egg after oviposition prevents adhesion retention ofeggs in the oviduct where they are inevitably jostledwould probably preclude adhesion and further develop-ment This may be a moot point however because oviduc-tal fluids would also prevent chalking as long as the eggremained in the oviduct

Another potential constraint on extended egg retentionfor crocodilians and turtles is that much of the calciumused in development is mobilized from the shell (Packardand Packard 1984) If calcification of the shell of turtle andcrocodilian embryos were reduced as a means to enhancegas exchange in the oviduct then the later growth of theembryo would be limited Calcium does not limit earlydevelopment however and selection could presumablyincrease the amount of calcium stored in the yolk Never-theless the suite of physiological and morphological fea-tures in turtles crocodilians and sphenodontidans mayform a substantial barrier to appreciable embryonic devel-opment in the oviducts for these reptiles

Squamates Oviposition at early developmental stagesmay explain why no crocodilian turtle or sphenodonti-dan has evolved viviparity (Shine 1983) On the otherhand squamate reptiles exhibit nearly the entire gamut ofpossible embryonic stages at oviposition At one extremeapproximately 20 of squamates are viviparous Close tothis extreme some oviparous species lay eggs with late-stage embryos For example the North American greensnake (Liochlorophis vernalis) the New Guinean skink(Sphenomorphus fragilis) and the Australian skink(Saiphos equalis) oviposit when embryos are within weeksor even days of hatching (Sexton and Claypool 1978Guillette 1992 Smith and Shine 1997) At the otherextreme oviposition when embryos are at the gastrulastage occurs in Chamaeleo chamaeleon (Bons and Bons1960) and at comparably early stages for varanid lizards(John Phillips Zoological Society of San Diego unpub-

lished data) A teiid Cnemidophorus uniparens ovipositsat stages 21ndash22 (Billy 1988) and several other species ofCnemidophorus oviposit considerably earlier (NormaManriacutequez Universidad Nacional Autoacutenoma de Meacutexicounpublished data) Reports of what appear to be extraor-dinarily long incubation periods after oviposition suggestthat the eggs of some chameleons may parallel those ofturtles not only in the stage at oviposition (ie gastrula-tion) but also in the potential for embryos to remain in anarrested state even after oviposition (Bons and Bons 1960Ewert 1985)

Despite the wide range of stages over which squamatesoviposit oviposition occurs much more frequently atsome stages than at others (Figure 3) Stages at ovipositiongroup tightly around a mode of stage 30 and the majori-ty of oviparous squamate taxa oviposit at stages 26ndash33This distribution suggests that the stage at oviposition isconstrained One possible constraint is on how early eggscan be laid Perhaps oviposition does not occur muchbefore stage 30 because of the time required to deposit theoviductal proteins (albumen) and the various layers ofshell that surround the ovum In chickens for exampleoviposition occurs 22 hours after ovulation by which timethe embryo has reached the gastrula stage Because reptilesare ectothermic their eggs are likely to require a longerperiod in the oviduct than those of birdsmdashand theirembryos will have attained more advanced embryonicstagesmdashbefore the shell is laid down

For example female Anolis carolinensis and Anolis lim-ifrons lay a sequence of one-egg clutches at minimumintervals of 6ndash7 days (Andrews 1979 1985) and oviposit atstage 30 (Robin M Andrews unpublished data) Ova arereleased into the right and left oviduct alternately suchthat females never have more than one egg per oviductand few females have an egg in both oviducts Thus eggshave a transit time of approximately 6 days (assuming thatno females have two oviductal eggs) to 9 days (assumingthat 50 of the females have two oviductal eggs simulta-neously) Similarly in the laboratory at a body tempera-ture of 30 degC Sceloporus woodi females secrete the pro-teinaceous fibers of the eggshell within the first 24 hoursafter ovulation calcium deposition is maximal on days3ndash9 but continues up to the time of oviposition whenembryos are at stage 27 and have been in the oviduct12ndash14 days (DeMarco 1993 Palmer et al 1993) These andother data (Miller 1985) on sea turtles suggest that reptileeggs have a necessary residence time in the oviduct of 1ndash2weeks before the albumen and shell are deposited

The critical question with respect to what causes ovipo-sition to occur in the vicinity of stage 30 is does oviposi-tion by squamates occur as soon as the shell is completeOr are eggs retained longer Oviposition at stage 30 corre-sponds to retention in the oviduct for 25 (DeMarco1993) to approximately 40 (Shine 1983) of total devel-opmental time (time in utero plus time in nest) Thesepercentages correspond to an egg retention time of 2ndash3

230 BioScience March 2000 Vol 50 No 3

Articles

weeks for small lizards that oviposit at stage 30 and thathave a total developmental time of 50ndash60 days Within aperiod of 2ndash3 weeks embryos will have reached stage 30during a period sufficient to completely shell the eggMany squamates that oviposit at stage 30 however havetotal developmental times that are longer than 50ndash60 days(Shine 1983 DeMarco 1993) If oviposition occurs at25ndash40 of the total developmental period eggs of suchspecies would be in the oviduct for considerably longerthan the 1ndash2 weeks required to completely shell the egg

This line of reasoning suggests that in species thatoviposit earlier than stage 30 either oviposition occurs assoon as eggs are shelled or embryonic development wasarrested before oviposition That is if development fromovulation to oviposition continued normally in theoviduct then after 2 weeks or more well-developedembryos (ie at approximately stage 30) should be pre-sent rather than the gastrula-stage embryos typical ofvaranids and some chameleons For example the periodbetween ovulation and oviposition for varanid lizards isconsistently approximately 1 month (Phillips and Millar1998) despite oviposition at very early stages of develop-ment By contrast squamates that oviposit at stage 30 orlater probably retain eggs for longer than the minimumperiod needed for deposition of the shell Designation ofstage 30 as the benchmark for judging how early eggs arelaid relative to shell formation is of course a crudeapproximation features of the shell or the body tempera-ture of the female could alter the length of time requiredto shell eggs

Another possible constraint that could cause oviposi-tion to cluster around stage 30 is how late in developmenteggs can be laid Such a constraint is particularly impor-tant because it directly limits the evolution of viviparityOnly a few oviparous species retain eggs well into the sec-ond half of development (Figure 3) which suggests thatthe shift from oviparity to viviparity does not occur oftenis rapid or both (Blackburn 1995) In any case the pauci-ty of species that oviposit at stages 34ndash39 suggests thatsuch a strategy presents problems for embryos the repro-ductive female or both

Embryos of oviparous species may experience a numberof problems if they are retained past stages 32ndash33mdasheventhose of species that normally oviposit at these stages

Large absolute increases in the mass of embryos in the sec-ond half of development are associated with greatlyenhanced demands for gas exchange and water (Black etal 1984 Birchard et al 1995) If these demands are notmet development of retained embryos may be retarded(Andrews and Rose 1994) Further retention of eggs at thispoint would be beneficial only if it ldquobought timerdquo for thefemale to locate a suitable oviposition site Such a physio-logical constraint may in addition to having specific con-sequences on embryonic development (see below) help toexplain why so few species lay eggs with embryos beyondstages 32ndash33

Females may also incur problems from extended eggretention The burden of the clutch would not changeappreciably during the first half of development but itwould during the second half when embryos undergolarge absolute increases in mass (Figure 2) as water is tak-en up by the extra-embryonic compartments of eggs and

March 2000 Vol 50 No 3 BioScience 231

Articles

Figure 3 Embryonic stage at oviposition for lizards (top)Lizards that lay a single clutch per reproductive season

(middle) Lizards that lay two or more clutches perreproductive season (bottom) All lizards The figures do notinclude varanids or the three species of Cnemidophorus for

which oviposition occurs before any visible development of theembryo (see text for details) The data are from a literature

review by Shine (1983) and an update to this data set byBlackburn (1995) with more recent data on stage at

oviposition and clutch frequency from Robin M Andrews(unpublished data)

the tissues of the embryo Although retained eggs may takeup less water than oviposited eggs (Mathies and Andrews1996) a gravid female that retains eggs into the secondhalf of development is faced not only with the problem oftransporting a clutch that is increasing in mass but alsowith the potential problem of passing enlarged eggsthrough the pelvic girdle during oviposition (Sinervo andLicht 1991) The rarity of oviposition after stage 34 maytherefore be a direct consequence of increasing egg size

Our concern in this article is largely with the proximatephysiological factors that constrain the embryonic stagesat which eggs are oviposited However it is also importantto recognize that many ultimate factors affect the stage atoviposition For example extended egg retention mayhave a high survival cost for some species because of theburden that the clutch imposes on the female (Sinervo andAdolph 1991) Such a cost would favor oviposition as soonas the shell is formed (as long as suitable nest sites areavailable) Clutch frequency should also affect the stage atoviposition Species that produce multiple clutches perseason could minimize the interval between clutches bylaying eggs at early stages of development thus producingmore clutches within a breeding season On the otherhand for species that are limited to one clutch per seasonthe length of time that eggs spend in the oviduct wouldnot reduce the reproductive output of the female Ifembryos continue to develop in the oviduct in suchspecies then delayed oviposition would shorten the lengthof time that the eggs are in the nest and vulnerable topredators or other mortality agents Single-clutchedspecies would thus be more likely than multiple-clutchedspecies to oviposit when embryos are at advanced stages

This prediction is borne out by data on lizards (Figure3) few multiple-clutched lizard species oviposit later thanstage 30 whereas most single-clutched lizard speciesoviposit at stage 30 or later (X2 = 70 P lt 001 2 acute 2 con-tingency analysis two-tailed test) Despite the associationbetween clutch frequency and stage at oviposition webelieve that selection for oviposition to occur at very earlyor very late stages of embryonic development will seldombe effective because of the time necessary to complete theegg shell on the one hand and physiological constraintson embryonic development in utero on the other

Consequences of extended egg retentionfor sceloporine lizardsOur research focuses on spiny lizards of the genus Scelo-porus and tree lizards of the genus Urosaurus Sceloporusand Urosaurus are sister taxa that is each otherrsquos closestrelative Urosaurus contains approximately 10 species allof which are oviparous Sceloporus contains approximate-ly 70 species almost half of which are viviparous (Sites etal 1992) Viviparity has apparently arisen only four timesin this genus once in the ancestor of the formosus grouponce in the ancestor of the grammicus megalepidurus andtorquatus groups and twice within the scalaris group

(Meacutendez-de la Cruz et al 1998) The scalaris group is theonly one to have both oviparous and viviparous speciesand the close relationship of the oviparous and viviparousspecies suggests that viviparity is of recent origin in thisgroup (Benabib et al 1997) The few origins of viviparityand the large number of species groups (approximately15) in which all members are oviparous suggests that thetransition to viviparity may be difficult in Sceloporus

Because oviposition at successively later stages ofembryonic development is the putative transition toviviparity the present-day consequences of extended eggretention on embryonic development provide insightsinto how easy it is for viviparity to evolve We have there-fore studied the consequences of the retention of eggsbeyond the normal stage of oviposition We initiallyfocused on oviparous species known to retain eggs toadvanced stages but a more complete picture requiredthat we also look at the consequences of egg retention bythe more typical speciesmdashthat is those that oviposit whenembryos reach approximately stage 30

The experimental protocol common to all of our stud-ies has been to keep gravid females of each species in thelaboratory under conditions that inhibit oviposition Thisapproach allowed us to assess the capacity of females tosupport embryonic development in utero past the time ofnormal oviposition To inhibit oviposition we tookadvantage of a normal response to drought because suc-cessful nesting in the field is dependent on adequate soilmoisture and rainfall is often unpredictable femalesrespond to drought by retaining their eggs (Andrews andRose 1994) Experimental females were thus kept in cageswith dry soil as a substrate which simulated drought con-ditions and inhibited oviposition (Andrews and Rose1994) Control females were housed under the same con-ditions as experimental females except that the cage sub-strate was moist soil most females oviposited normallyunder these conditions Eggs from control females wereincubated in a standard incubation medium and at thesame temperature as eggs within experimental females

One set of experiments examined egg retention bySceloporus scalaris an oviparous species that is a model forthe transition from oviparity to viviparity (Figure 4) Sscalaris a member of the scalaris species group occurs ingrasslands at many elevations in Arizona New Mexicoand northern Mexico Females from a low-elevation pop-ulation are able to retain eggs for an unprecedented lengthof time although they normally oviposit at stages 31ndash33(Mathies and Andrews 1995 1996 Thomas Mathies andRobin M Andrews unpublished data) Eggs retained aslong as 1 month (to stage 38) and surgically removed fromfemales develop at the same rate and have the same sur-vival to hatching rate as control eggs Eggs retained for afurther 10ndash13 days advance to stage 40 these eggs pro-duced viable hatchlings 1ndash3 days after they were surgicallyremoved from females When the same experiment wasconducted at a slightly higher incubation temperature

232 BioScience March 2000 Vol 50 No 3

Articles

development was somewhat slowerfor retained embryos than for con-trol embryos presumably becausethe higher incubation temperatureincreased the oxygen requirementsof embryos beyond that availablein utero (Andrews 1997)

Nevertheless among squamatesS scalaris has an unusual ability toextend egg retention to within afew days of hatching and to do sowith little or no impairment ofembryonic development The abil-ity to support embryogenesis inutero to advanced stages is alsofound in another oviparous speciesin the scalaris group Sceloporusaeneus (Robin M Andrewsunpublished observation) the sis-ter species of the viviparous Scelo-porus bicanthalis The ability ofoviparous species in the scalarisgroup to extend egg retention toadvanced embryonic stages has likely facilitated the evolu-tion of viviparity in this group

Sceloporus virgatus an oviparous member of the undu-latus species group also normally lays eggs at stages 31ndash33In contrast to S scalaris however when eggs of S virgatusare retained past the normal time of oviposition as long asto stage 37 embryonic development is retarded (Figure 4Andrews and Rose 1994 Andrews 1997) For example thedry masses of embryos retained for 30 days past the nor-mal time of oviposition were less than half those of con-trols and retained embryos showed considerably lowersurvival to hatching than control eggs Female S virgatusthat retain eggs beyond the normal time of ovipositionthus have greatly reduced fitness This assertion is not toimply that extended egg retention has no benefit in thisspeciesmdashfemales that do not retain eggs until the seasonalrains occur would have zero fitness for that reproductiveseason

S virgatus appears to be unusual both within the undu-latus species group and within the genus Sceloporus in itsability to normally support embryonic development paststage 30 (Mathies 1998) No other species that we studiedin the undulatus species group is able to retain eggs appre-ciably beyond stage 30 For example females of Sceloporusoccidentalis another member of the undulatus groupoviposit when embryos reach stages 28ndash30 even if condi-tions are not suitable for nesting This pattern is also char-acteristic of Sceloporus graciosus (a member of the gracio-sus species group) Females of two members of theundulatus species groupmdashSceloporus undulatus consobri-nus from Arizona and Sceloporus undulatus hyacinthinusfrom Virginia and Missourimdashboth typically oviposit whenembryos are at stage 28ndash30 (Sexton and Marion 1974

Mathies 1998) as do female Sceloporus clarki (a member ofthe clarki species group) Moreover embryos of thesespecies do not develop further when eggs are retainedbeyond the normal time of oviposition (Mathies 1998)Initially the similar stages at oviposition for S u consobri-nus and S u hyacinthinus housed on moist and dry sub-strates would appear to be the result of the brief periodapproximately 10 days of facultative egg retention underdrought conditions However during this 10-day periodembryos in control eggs had reached stage 32 and had drymasses four to eight times greater than those of theretained embryos These observations indicate that devel-opment of retained embryos was arrested Females in theundulatus graciosus and clarki species groups thus exhib-it a limited ability to extend egg retention and to supportembryogenesis in utero beyond stage 30

In contrast to members of the undulatus group a mem-ber of the sister genus to Sceloporus Urosaurus ornatus isable to retain eggs for at least a month past the time ofnormal oviposition (Mathies and Andrews 1999) As inmost members of the undulatus group however develop-ment is arrested at approximately the time of normal ovipo-sition In this case development is arrested at stage 305whereas control embryos reach stage 38 after the samelength of incubation (Figure 4) When retained eggs arelaid development is reinitiated and continues to hatchingHatching success of retained eggs is as high as that of con-trol eggs Retention in Urosaurus thus has no obvious effecton embryonic development other than to delay hatchingand then only by the time spent in arrest Retention doesaffect hatchlings however Hatchlings produced after longperiods of retention are smaller and less hydrated thanhatchlings produced after short periods of retention

March 2000 Vol 50 No 3 BioScience 233

Articles

Figure 4 Embryo stage in three species of phrynosomatid lizards as a function of dayselapsed since the normal time of oviposition Eggs of control females (open circles)were oviposited at the normal time Eggs of experimental females (solid circles) wereretained in the oviducts for the number of days indicated Oviposition byexperimental females was inhibited by providing dry substrates for nesting Thedevelopment of the retained embryos was arrested for Urosaurus ornatus (left) wasretarded for Sceloporus virgatus (middle) and proceeded normally for Sceloporusscalaris (right) Figure modified from Andrews and Rose (1994) Mathies and Andrews(1996) and Mathies (1998)

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

focuses on the consequences of extended egg retention onembryonic development in oviparous species and thusidentifies potential constraints on the shift from oviparityto viviparity (more precisely lecithotrophy sensu Black-burn 1992) This approach lends itself to understandingthe evolution of viviparity at high taxonomic levels (egturtles versus squamates) as well as at low taxonomic lev-els (within genera and species)

We begin by reviewing the general pattern of embryon-ic development of reptiles to set the framework for further

discussion We then summarize the timing of ovipositionin reptiles and how timing affects embryonic developmentin the oviducts If as we assume the transition betweenoviparity and viviparity involves a progressive increase inthe amount of development that takes place in theoviduct then any factor that limits development in theoviduct also limits the potential for viviparity to evolveWe therefore discuss a number of reasons why ovipositionoccurs at particular stages of embryonic development Wenext consider the consequences of extended egg retentionin a clade of phrynosomatid lizards and how these conse-quences may restrain or enhance the evolution of vivipar-ity in this group Finally we summarize evidence thathypoxia (oxygen deprivation) is the physiological factorthat limits development of embryos in the oviducts

Comparative embryologydevelopmental chronologyEmbryonic development is a continuum from fertilizationto hatching or birth For comparative purposes howeverembryologists break development into a series of discretestages that are demarcated by the appearance of distinctfeatures such as neurulation torsion pharyngeal cleftslimb buds and so on Ideally a ldquonormal tablerdquo of develop-ment would list these stages extending from fertilizationto hatching or birth and would report the timing of eventsat some specific temperature Such an ideal normal tableis however not available for any reptile For example themost commonly used normal table for squamates starts atcleavage and has 40 stages but these stages are not relatedto time (Dufaure and Hubert 1961) In contrast whereasnormal tables for crocodilians provide the timing of devel-opmental events for particular incubation temperaturesstaging does not start until oviposition (neurulation) andthe normal table has only 28 stages (Ferguson 1985)

At a general level however patterns of development arecomparable among reptilian taxa First differentiation israpid initially but slows appreciably as the embryoapproaches the end of development with embryos passingthrough 80 or more of their recognized stages byapproximately halfway through the developmental periodThe relationship between developmental stage and timefor crocodilian and sea turtle embryos illustrates this phe-nomenon (Figure 1) The finding that most stages arecompleted before mid-development irrespective of taxonreflects the fact that early development is characterized bytissue differentiation and organogenesis processes thatprovide many distinct landmarks and thus result in theidentification of many distinct stages Later developmentis largely characterized by growth in size and by biochem-ical and physiological changes that provide few new fea-tures for visual staging Indeed by mid-developmentembryos are small but recognizable lizards snakes croco-diles or turtles

A second general pattern of reptilian development isthat growth in size is exponential (Figure 2) The magni-

228 BioScience March 2000 Vol 50 No 3

Articles

Figure 1 The relationship between embryonic stage and timesince oviposition with time expressed as percentage of theincubation period in crocodilian and sea turtle embryos(top) For Crocodylus porosus (Ferguson 1985) staging startsat oviposition when the embryo is at the neurula stage(roughly equivalent to stage 14 for marine turtles) (bottom)For marine turtles (Miller 1985) staging starts at ovipositionwhen the embryo is at the gastrula stage The time betweenovulation and oviposition is estimated (Miller 1985) to be 8ndash9daysmdashequivalent to 14 of the time between oviposition andhatching (60 days) The shapes of the curves for both reptilesare similar rapid development during the first half ofincubation reflects tissue and organ differentiation andslower development later in development reflects an increasein embryo size and biochemical and physiological changesthat provide few new features for visual staging

tude of growth toward the end of development is consid-erably greater than depicted in Figure 2 for two reasonssize is indexed by length rather than mass and the x-axisrepresents stage rather than time That is if the axes weremass and time the curve would exhibit a much greaterincrease in size after mid-development than shown in Fig-ure 2 The second half of development is thus character-ized by a size increase of considerable magnitude

Stage at ovipositionFor oviparous reptiles oviposition marks the transitionbetween embryonic development within the mother and inthe environment The timing of this transition varies con-siderably among reptilian groups Eggs of turtles crocodil-ians and sphenodontidans are invariably laid whenembryos are in the earliest stages of development In con-trast the eggs of most squamates are laid when embryos areapproximately one-third through development althoughoviposition can occur substantially earlier or later As wediscuss several morphological and physiological factorsmay constrain the stage at oviposition and thus account forvariation in the reproductive modes of reptiles

Turtles crocodilians and sphenodontidans Ovi-position by turtles (Chelonia) and tuataras (Sphenodonti-da) occurs when the embryo is at the gastrula stage hencelittle development occurs in the oviduct (Ewert 1985 Mof-fat 1985) After reaching the gastrula stage turtle embryosenter developmental arrest resuming development onlyafter oviposition The amount of time that turtle embryosspend in developmental arrest varies Turtle eggs remainin the oviduct for at least the 2-week period that is neces-sary for shell formation However the length of egg reten-tion can vary considerably among females The longestperiods of egg retention for the North American sliderturtle Trachemys scripta and the mud turtle Kinosternonsubrubrum in South Carolina were 39 days and 50 daysrespectively (Buhlmann et al 1995) Some females of thechicken turtle Deirochelys reticularia retain their eggsover winter and lay them the following spring presumablybecause conditions in the fall were not favorable for suc-cessful nesting These eggs are thus retained by the femalefor 4ndash7 months before oviposition (Buhlmann et al 1995)and hatch in the spring of the following year (Congdon etal 1983) In contrast to all other reptiles a long period ofegg retention is characteristic of tuataras eggs are in theoviduct for 6ndash8 months before oviposition (Cree et al1992) Development is presumably arrested while eggs arein the oviduct because embryos are at only the gastrulastage at oviposition

Crocodilians oviposit when embryos are at the neurulastage and thus somewhat more advanced than those ofturtles at the time of oviposition (Ferguson 1985) Unliketurtle embryos however crocodilian embryos do notundergo developmental arrest and therefore must be laidimmediately Nevertheless oviposition at early develop-

mental stages appears to be characteristic of all turtlescrocodilians and sphenodontidans The fact that embryosof these taxa do not develop beyond an early developmen-tal stage while in the oviduct has apparently limited thereproductive options open to these taxamdashno turtle croc-odilian or sphenodontidan is viviparous

For turtles crocodilians and sphenodontidans theapparent requirement for oviposition to occur early indevelopment may reflect a suite of morphological andphysiological constraints that were established early in thehistory of these taxa The most immediate constraint onembryonic development may be the limited exchange ofgases particularly oxygen in the oviduct Gas diffusion isconsiderably slower in water than in air and the pores andchannels of the eggshell are filled with fluid when the eggis in the oviduct Moreover the heavily calcified shells ofthe eggs of crocodilians and most turtles (Packard andDeMarco 1991) may further exacerbate the problem of gasexchange Indirect evidence to support this hypothesis forturtles includes the observations that the shell calcifies justbefore gastrulation (when development becomes arrested)and that after oviposition embryos of species with morepermeable flexible-shelled eggs develop faster thanembryos of species with rigid calcareous shells (Ewert1985)

Experimental studies on the northern snake-necked

March 2000 Vol 50 No 3 BioScience 229

Articles

Figure 2 The relationship between embryo size and stagefor Lacerta vivipara a viviparous lizard Embryo size isindexed by its length which actually underestimates theactual size of the embryo (see text for details) Embryo stagefollows Dufaure and Hubert (1961) Stages at whichimportant developmental events occur (gastrulationneurulation appearance of limb buds and digitscompletely differentiated) the range of stages over whichoviposition occurs in the majority of oviparous squamatesand the stage of parturition in viviparous squamates areindicated Figure modified from Xavier and Gavaud(1986)

turtle Chelodina rugosa in Australia provide direct evi-dence that hypoxia maintains developmental arrest in thisspecies (Kennett et al 1993) The rigid-shelled eggs arelaid underwater in seasonal ponds and the embryosremain in developmental arrest as long as the eggs areimmersed in water Embryos resume development onlyafter the soil dries and oxygen tension rises In an experi-ment in which eggs were shifted from water to an atmos-phere of pure nitrogen embryos remained in develop-mental arrest they resumed development only whenexposed to atmospheric air

An alternative or perhaps complementary explanationfor oviposition early in development by turtles and croco-dilians concerns the adhesion of the embryo and thevitelline membrane to the shell (Ewert 1985 Ferguson1985) Adhesion shortly after oviposition appears to havea respiratory function because the ldquochalkingrdquo (drying) ofthe shell that is associated with adhesion increases its con-ductance to gases (Thompson 1985) Because movementof an egg after oviposition prevents adhesion retention ofeggs in the oviduct where they are inevitably jostledwould probably preclude adhesion and further develop-ment This may be a moot point however because oviduc-tal fluids would also prevent chalking as long as the eggremained in the oviduct

Another potential constraint on extended egg retentionfor crocodilians and turtles is that much of the calciumused in development is mobilized from the shell (Packardand Packard 1984) If calcification of the shell of turtle andcrocodilian embryos were reduced as a means to enhancegas exchange in the oviduct then the later growth of theembryo would be limited Calcium does not limit earlydevelopment however and selection could presumablyincrease the amount of calcium stored in the yolk Never-theless the suite of physiological and morphological fea-tures in turtles crocodilians and sphenodontidans mayform a substantial barrier to appreciable embryonic devel-opment in the oviducts for these reptiles

Squamates Oviposition at early developmental stagesmay explain why no crocodilian turtle or sphenodonti-dan has evolved viviparity (Shine 1983) On the otherhand squamate reptiles exhibit nearly the entire gamut ofpossible embryonic stages at oviposition At one extremeapproximately 20 of squamates are viviparous Close tothis extreme some oviparous species lay eggs with late-stage embryos For example the North American greensnake (Liochlorophis vernalis) the New Guinean skink(Sphenomorphus fragilis) and the Australian skink(Saiphos equalis) oviposit when embryos are within weeksor even days of hatching (Sexton and Claypool 1978Guillette 1992 Smith and Shine 1997) At the otherextreme oviposition when embryos are at the gastrulastage occurs in Chamaeleo chamaeleon (Bons and Bons1960) and at comparably early stages for varanid lizards(John Phillips Zoological Society of San Diego unpub-

lished data) A teiid Cnemidophorus uniparens ovipositsat stages 21ndash22 (Billy 1988) and several other species ofCnemidophorus oviposit considerably earlier (NormaManriacutequez Universidad Nacional Autoacutenoma de Meacutexicounpublished data) Reports of what appear to be extraor-dinarily long incubation periods after oviposition suggestthat the eggs of some chameleons may parallel those ofturtles not only in the stage at oviposition (ie gastrula-tion) but also in the potential for embryos to remain in anarrested state even after oviposition (Bons and Bons 1960Ewert 1985)

Despite the wide range of stages over which squamatesoviposit oviposition occurs much more frequently atsome stages than at others (Figure 3) Stages at ovipositiongroup tightly around a mode of stage 30 and the majori-ty of oviparous squamate taxa oviposit at stages 26ndash33This distribution suggests that the stage at oviposition isconstrained One possible constraint is on how early eggscan be laid Perhaps oviposition does not occur muchbefore stage 30 because of the time required to deposit theoviductal proteins (albumen) and the various layers ofshell that surround the ovum In chickens for exampleoviposition occurs 22 hours after ovulation by which timethe embryo has reached the gastrula stage Because reptilesare ectothermic their eggs are likely to require a longerperiod in the oviduct than those of birdsmdashand theirembryos will have attained more advanced embryonicstagesmdashbefore the shell is laid down

For example female Anolis carolinensis and Anolis lim-ifrons lay a sequence of one-egg clutches at minimumintervals of 6ndash7 days (Andrews 1979 1985) and oviposit atstage 30 (Robin M Andrews unpublished data) Ova arereleased into the right and left oviduct alternately suchthat females never have more than one egg per oviductand few females have an egg in both oviducts Thus eggshave a transit time of approximately 6 days (assuming thatno females have two oviductal eggs) to 9 days (assumingthat 50 of the females have two oviductal eggs simulta-neously) Similarly in the laboratory at a body tempera-ture of 30 degC Sceloporus woodi females secrete the pro-teinaceous fibers of the eggshell within the first 24 hoursafter ovulation calcium deposition is maximal on days3ndash9 but continues up to the time of oviposition whenembryos are at stage 27 and have been in the oviduct12ndash14 days (DeMarco 1993 Palmer et al 1993) These andother data (Miller 1985) on sea turtles suggest that reptileeggs have a necessary residence time in the oviduct of 1ndash2weeks before the albumen and shell are deposited

The critical question with respect to what causes ovipo-sition to occur in the vicinity of stage 30 is does oviposi-tion by squamates occur as soon as the shell is completeOr are eggs retained longer Oviposition at stage 30 corre-sponds to retention in the oviduct for 25 (DeMarco1993) to approximately 40 (Shine 1983) of total devel-opmental time (time in utero plus time in nest) Thesepercentages correspond to an egg retention time of 2ndash3

230 BioScience March 2000 Vol 50 No 3

Articles

weeks for small lizards that oviposit at stage 30 and thathave a total developmental time of 50ndash60 days Within aperiod of 2ndash3 weeks embryos will have reached stage 30during a period sufficient to completely shell the eggMany squamates that oviposit at stage 30 however havetotal developmental times that are longer than 50ndash60 days(Shine 1983 DeMarco 1993) If oviposition occurs at25ndash40 of the total developmental period eggs of suchspecies would be in the oviduct for considerably longerthan the 1ndash2 weeks required to completely shell the egg

This line of reasoning suggests that in species thatoviposit earlier than stage 30 either oviposition occurs assoon as eggs are shelled or embryonic development wasarrested before oviposition That is if development fromovulation to oviposition continued normally in theoviduct then after 2 weeks or more well-developedembryos (ie at approximately stage 30) should be pre-sent rather than the gastrula-stage embryos typical ofvaranids and some chameleons For example the periodbetween ovulation and oviposition for varanid lizards isconsistently approximately 1 month (Phillips and Millar1998) despite oviposition at very early stages of develop-ment By contrast squamates that oviposit at stage 30 orlater probably retain eggs for longer than the minimumperiod needed for deposition of the shell Designation ofstage 30 as the benchmark for judging how early eggs arelaid relative to shell formation is of course a crudeapproximation features of the shell or the body tempera-ture of the female could alter the length of time requiredto shell eggs

Another possible constraint that could cause oviposi-tion to cluster around stage 30 is how late in developmenteggs can be laid Such a constraint is particularly impor-tant because it directly limits the evolution of viviparityOnly a few oviparous species retain eggs well into the sec-ond half of development (Figure 3) which suggests thatthe shift from oviparity to viviparity does not occur oftenis rapid or both (Blackburn 1995) In any case the pauci-ty of species that oviposit at stages 34ndash39 suggests thatsuch a strategy presents problems for embryos the repro-ductive female or both

Embryos of oviparous species may experience a numberof problems if they are retained past stages 32ndash33mdasheventhose of species that normally oviposit at these stages

Large absolute increases in the mass of embryos in the sec-ond half of development are associated with greatlyenhanced demands for gas exchange and water (Black etal 1984 Birchard et al 1995) If these demands are notmet development of retained embryos may be retarded(Andrews and Rose 1994) Further retention of eggs at thispoint would be beneficial only if it ldquobought timerdquo for thefemale to locate a suitable oviposition site Such a physio-logical constraint may in addition to having specific con-sequences on embryonic development (see below) help toexplain why so few species lay eggs with embryos beyondstages 32ndash33

Females may also incur problems from extended eggretention The burden of the clutch would not changeappreciably during the first half of development but itwould during the second half when embryos undergolarge absolute increases in mass (Figure 2) as water is tak-en up by the extra-embryonic compartments of eggs and

March 2000 Vol 50 No 3 BioScience 231

Articles

Figure 3 Embryonic stage at oviposition for lizards (top)Lizards that lay a single clutch per reproductive season

(middle) Lizards that lay two or more clutches perreproductive season (bottom) All lizards The figures do notinclude varanids or the three species of Cnemidophorus for

which oviposition occurs before any visible development of theembryo (see text for details) The data are from a literature

review by Shine (1983) and an update to this data set byBlackburn (1995) with more recent data on stage at

oviposition and clutch frequency from Robin M Andrews(unpublished data)

the tissues of the embryo Although retained eggs may takeup less water than oviposited eggs (Mathies and Andrews1996) a gravid female that retains eggs into the secondhalf of development is faced not only with the problem oftransporting a clutch that is increasing in mass but alsowith the potential problem of passing enlarged eggsthrough the pelvic girdle during oviposition (Sinervo andLicht 1991) The rarity of oviposition after stage 34 maytherefore be a direct consequence of increasing egg size

Our concern in this article is largely with the proximatephysiological factors that constrain the embryonic stagesat which eggs are oviposited However it is also importantto recognize that many ultimate factors affect the stage atoviposition For example extended egg retention mayhave a high survival cost for some species because of theburden that the clutch imposes on the female (Sinervo andAdolph 1991) Such a cost would favor oviposition as soonas the shell is formed (as long as suitable nest sites areavailable) Clutch frequency should also affect the stage atoviposition Species that produce multiple clutches perseason could minimize the interval between clutches bylaying eggs at early stages of development thus producingmore clutches within a breeding season On the otherhand for species that are limited to one clutch per seasonthe length of time that eggs spend in the oviduct wouldnot reduce the reproductive output of the female Ifembryos continue to develop in the oviduct in suchspecies then delayed oviposition would shorten the lengthof time that the eggs are in the nest and vulnerable topredators or other mortality agents Single-clutchedspecies would thus be more likely than multiple-clutchedspecies to oviposit when embryos are at advanced stages

This prediction is borne out by data on lizards (Figure3) few multiple-clutched lizard species oviposit later thanstage 30 whereas most single-clutched lizard speciesoviposit at stage 30 or later (X2 = 70 P lt 001 2 acute 2 con-tingency analysis two-tailed test) Despite the associationbetween clutch frequency and stage at oviposition webelieve that selection for oviposition to occur at very earlyor very late stages of embryonic development will seldombe effective because of the time necessary to complete theegg shell on the one hand and physiological constraintson embryonic development in utero on the other

Consequences of extended egg retentionfor sceloporine lizardsOur research focuses on spiny lizards of the genus Scelo-porus and tree lizards of the genus Urosaurus Sceloporusand Urosaurus are sister taxa that is each otherrsquos closestrelative Urosaurus contains approximately 10 species allof which are oviparous Sceloporus contains approximate-ly 70 species almost half of which are viviparous (Sites etal 1992) Viviparity has apparently arisen only four timesin this genus once in the ancestor of the formosus grouponce in the ancestor of the grammicus megalepidurus andtorquatus groups and twice within the scalaris group

(Meacutendez-de la Cruz et al 1998) The scalaris group is theonly one to have both oviparous and viviparous speciesand the close relationship of the oviparous and viviparousspecies suggests that viviparity is of recent origin in thisgroup (Benabib et al 1997) The few origins of viviparityand the large number of species groups (approximately15) in which all members are oviparous suggests that thetransition to viviparity may be difficult in Sceloporus

Because oviposition at successively later stages ofembryonic development is the putative transition toviviparity the present-day consequences of extended eggretention on embryonic development provide insightsinto how easy it is for viviparity to evolve We have there-fore studied the consequences of the retention of eggsbeyond the normal stage of oviposition We initiallyfocused on oviparous species known to retain eggs toadvanced stages but a more complete picture requiredthat we also look at the consequences of egg retention bythe more typical speciesmdashthat is those that oviposit whenembryos reach approximately stage 30

The experimental protocol common to all of our stud-ies has been to keep gravid females of each species in thelaboratory under conditions that inhibit oviposition Thisapproach allowed us to assess the capacity of females tosupport embryonic development in utero past the time ofnormal oviposition To inhibit oviposition we tookadvantage of a normal response to drought because suc-cessful nesting in the field is dependent on adequate soilmoisture and rainfall is often unpredictable femalesrespond to drought by retaining their eggs (Andrews andRose 1994) Experimental females were thus kept in cageswith dry soil as a substrate which simulated drought con-ditions and inhibited oviposition (Andrews and Rose1994) Control females were housed under the same con-ditions as experimental females except that the cage sub-strate was moist soil most females oviposited normallyunder these conditions Eggs from control females wereincubated in a standard incubation medium and at thesame temperature as eggs within experimental females

One set of experiments examined egg retention bySceloporus scalaris an oviparous species that is a model forthe transition from oviparity to viviparity (Figure 4) Sscalaris a member of the scalaris species group occurs ingrasslands at many elevations in Arizona New Mexicoand northern Mexico Females from a low-elevation pop-ulation are able to retain eggs for an unprecedented lengthof time although they normally oviposit at stages 31ndash33(Mathies and Andrews 1995 1996 Thomas Mathies andRobin M Andrews unpublished data) Eggs retained aslong as 1 month (to stage 38) and surgically removed fromfemales develop at the same rate and have the same sur-vival to hatching rate as control eggs Eggs retained for afurther 10ndash13 days advance to stage 40 these eggs pro-duced viable hatchlings 1ndash3 days after they were surgicallyremoved from females When the same experiment wasconducted at a slightly higher incubation temperature

232 BioScience March 2000 Vol 50 No 3

Articles

development was somewhat slowerfor retained embryos than for con-trol embryos presumably becausethe higher incubation temperatureincreased the oxygen requirementsof embryos beyond that availablein utero (Andrews 1997)

Nevertheless among squamatesS scalaris has an unusual ability toextend egg retention to within afew days of hatching and to do sowith little or no impairment ofembryonic development The abil-ity to support embryogenesis inutero to advanced stages is alsofound in another oviparous speciesin the scalaris group Sceloporusaeneus (Robin M Andrewsunpublished observation) the sis-ter species of the viviparous Scelo-porus bicanthalis The ability ofoviparous species in the scalarisgroup to extend egg retention toadvanced embryonic stages has likely facilitated the evolu-tion of viviparity in this group

Sceloporus virgatus an oviparous member of the undu-latus species group also normally lays eggs at stages 31ndash33In contrast to S scalaris however when eggs of S virgatusare retained past the normal time of oviposition as long asto stage 37 embryonic development is retarded (Figure 4Andrews and Rose 1994 Andrews 1997) For example thedry masses of embryos retained for 30 days past the nor-mal time of oviposition were less than half those of con-trols and retained embryos showed considerably lowersurvival to hatching than control eggs Female S virgatusthat retain eggs beyond the normal time of ovipositionthus have greatly reduced fitness This assertion is not toimply that extended egg retention has no benefit in thisspeciesmdashfemales that do not retain eggs until the seasonalrains occur would have zero fitness for that reproductiveseason

S virgatus appears to be unusual both within the undu-latus species group and within the genus Sceloporus in itsability to normally support embryonic development paststage 30 (Mathies 1998) No other species that we studiedin the undulatus species group is able to retain eggs appre-ciably beyond stage 30 For example females of Sceloporusoccidentalis another member of the undulatus groupoviposit when embryos reach stages 28ndash30 even if condi-tions are not suitable for nesting This pattern is also char-acteristic of Sceloporus graciosus (a member of the gracio-sus species group) Females of two members of theundulatus species groupmdashSceloporus undulatus consobri-nus from Arizona and Sceloporus undulatus hyacinthinusfrom Virginia and Missourimdashboth typically oviposit whenembryos are at stage 28ndash30 (Sexton and Marion 1974

Mathies 1998) as do female Sceloporus clarki (a member ofthe clarki species group) Moreover embryos of thesespecies do not develop further when eggs are retainedbeyond the normal time of oviposition (Mathies 1998)Initially the similar stages at oviposition for S u consobri-nus and S u hyacinthinus housed on moist and dry sub-strates would appear to be the result of the brief periodapproximately 10 days of facultative egg retention underdrought conditions However during this 10-day periodembryos in control eggs had reached stage 32 and had drymasses four to eight times greater than those of theretained embryos These observations indicate that devel-opment of retained embryos was arrested Females in theundulatus graciosus and clarki species groups thus exhib-it a limited ability to extend egg retention and to supportembryogenesis in utero beyond stage 30

In contrast to members of the undulatus group a mem-ber of the sister genus to Sceloporus Urosaurus ornatus isable to retain eggs for at least a month past the time ofnormal oviposition (Mathies and Andrews 1999) As inmost members of the undulatus group however develop-ment is arrested at approximately the time of normal ovipo-sition In this case development is arrested at stage 305whereas control embryos reach stage 38 after the samelength of incubation (Figure 4) When retained eggs arelaid development is reinitiated and continues to hatchingHatching success of retained eggs is as high as that of con-trol eggs Retention in Urosaurus thus has no obvious effecton embryonic development other than to delay hatchingand then only by the time spent in arrest Retention doesaffect hatchlings however Hatchlings produced after longperiods of retention are smaller and less hydrated thanhatchlings produced after short periods of retention

March 2000 Vol 50 No 3 BioScience 233

Articles

Figure 4 Embryo stage in three species of phrynosomatid lizards as a function of dayselapsed since the normal time of oviposition Eggs of control females (open circles)were oviposited at the normal time Eggs of experimental females (solid circles) wereretained in the oviducts for the number of days indicated Oviposition byexperimental females was inhibited by providing dry substrates for nesting Thedevelopment of the retained embryos was arrested for Urosaurus ornatus (left) wasretarded for Sceloporus virgatus (middle) and proceeded normally for Sceloporusscalaris (right) Figure modified from Andrews and Rose (1994) Mathies and Andrews(1996) and Mathies (1998)

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

tude of growth toward the end of development is consid-erably greater than depicted in Figure 2 for two reasonssize is indexed by length rather than mass and the x-axisrepresents stage rather than time That is if the axes weremass and time the curve would exhibit a much greaterincrease in size after mid-development than shown in Fig-ure 2 The second half of development is thus character-ized by a size increase of considerable magnitude

Stage at ovipositionFor oviparous reptiles oviposition marks the transitionbetween embryonic development within the mother and inthe environment The timing of this transition varies con-siderably among reptilian groups Eggs of turtles crocodil-ians and sphenodontidans are invariably laid whenembryos are in the earliest stages of development In con-trast the eggs of most squamates are laid when embryos areapproximately one-third through development althoughoviposition can occur substantially earlier or later As wediscuss several morphological and physiological factorsmay constrain the stage at oviposition and thus account forvariation in the reproductive modes of reptiles

Turtles crocodilians and sphenodontidans Ovi-position by turtles (Chelonia) and tuataras (Sphenodonti-da) occurs when the embryo is at the gastrula stage hencelittle development occurs in the oviduct (Ewert 1985 Mof-fat 1985) After reaching the gastrula stage turtle embryosenter developmental arrest resuming development onlyafter oviposition The amount of time that turtle embryosspend in developmental arrest varies Turtle eggs remainin the oviduct for at least the 2-week period that is neces-sary for shell formation However the length of egg reten-tion can vary considerably among females The longestperiods of egg retention for the North American sliderturtle Trachemys scripta and the mud turtle Kinosternonsubrubrum in South Carolina were 39 days and 50 daysrespectively (Buhlmann et al 1995) Some females of thechicken turtle Deirochelys reticularia retain their eggsover winter and lay them the following spring presumablybecause conditions in the fall were not favorable for suc-cessful nesting These eggs are thus retained by the femalefor 4ndash7 months before oviposition (Buhlmann et al 1995)and hatch in the spring of the following year (Congdon etal 1983) In contrast to all other reptiles a long period ofegg retention is characteristic of tuataras eggs are in theoviduct for 6ndash8 months before oviposition (Cree et al1992) Development is presumably arrested while eggs arein the oviduct because embryos are at only the gastrulastage at oviposition

Crocodilians oviposit when embryos are at the neurulastage and thus somewhat more advanced than those ofturtles at the time of oviposition (Ferguson 1985) Unliketurtle embryos however crocodilian embryos do notundergo developmental arrest and therefore must be laidimmediately Nevertheless oviposition at early develop-

mental stages appears to be characteristic of all turtlescrocodilians and sphenodontidans The fact that embryosof these taxa do not develop beyond an early developmen-tal stage while in the oviduct has apparently limited thereproductive options open to these taxamdashno turtle croc-odilian or sphenodontidan is viviparous

For turtles crocodilians and sphenodontidans theapparent requirement for oviposition to occur early indevelopment may reflect a suite of morphological andphysiological constraints that were established early in thehistory of these taxa The most immediate constraint onembryonic development may be the limited exchange ofgases particularly oxygen in the oviduct Gas diffusion isconsiderably slower in water than in air and the pores andchannels of the eggshell are filled with fluid when the eggis in the oviduct Moreover the heavily calcified shells ofthe eggs of crocodilians and most turtles (Packard andDeMarco 1991) may further exacerbate the problem of gasexchange Indirect evidence to support this hypothesis forturtles includes the observations that the shell calcifies justbefore gastrulation (when development becomes arrested)and that after oviposition embryos of species with morepermeable flexible-shelled eggs develop faster thanembryos of species with rigid calcareous shells (Ewert1985)

Experimental studies on the northern snake-necked

March 2000 Vol 50 No 3 BioScience 229

Articles

Figure 2 The relationship between embryo size and stagefor Lacerta vivipara a viviparous lizard Embryo size isindexed by its length which actually underestimates theactual size of the embryo (see text for details) Embryo stagefollows Dufaure and Hubert (1961) Stages at whichimportant developmental events occur (gastrulationneurulation appearance of limb buds and digitscompletely differentiated) the range of stages over whichoviposition occurs in the majority of oviparous squamatesand the stage of parturition in viviparous squamates areindicated Figure modified from Xavier and Gavaud(1986)

turtle Chelodina rugosa in Australia provide direct evi-dence that hypoxia maintains developmental arrest in thisspecies (Kennett et al 1993) The rigid-shelled eggs arelaid underwater in seasonal ponds and the embryosremain in developmental arrest as long as the eggs areimmersed in water Embryos resume development onlyafter the soil dries and oxygen tension rises In an experi-ment in which eggs were shifted from water to an atmos-phere of pure nitrogen embryos remained in develop-mental arrest they resumed development only whenexposed to atmospheric air

An alternative or perhaps complementary explanationfor oviposition early in development by turtles and croco-dilians concerns the adhesion of the embryo and thevitelline membrane to the shell (Ewert 1985 Ferguson1985) Adhesion shortly after oviposition appears to havea respiratory function because the ldquochalkingrdquo (drying) ofthe shell that is associated with adhesion increases its con-ductance to gases (Thompson 1985) Because movementof an egg after oviposition prevents adhesion retention ofeggs in the oviduct where they are inevitably jostledwould probably preclude adhesion and further develop-ment This may be a moot point however because oviduc-tal fluids would also prevent chalking as long as the eggremained in the oviduct

Another potential constraint on extended egg retentionfor crocodilians and turtles is that much of the calciumused in development is mobilized from the shell (Packardand Packard 1984) If calcification of the shell of turtle andcrocodilian embryos were reduced as a means to enhancegas exchange in the oviduct then the later growth of theembryo would be limited Calcium does not limit earlydevelopment however and selection could presumablyincrease the amount of calcium stored in the yolk Never-theless the suite of physiological and morphological fea-tures in turtles crocodilians and sphenodontidans mayform a substantial barrier to appreciable embryonic devel-opment in the oviducts for these reptiles

Squamates Oviposition at early developmental stagesmay explain why no crocodilian turtle or sphenodonti-dan has evolved viviparity (Shine 1983) On the otherhand squamate reptiles exhibit nearly the entire gamut ofpossible embryonic stages at oviposition At one extremeapproximately 20 of squamates are viviparous Close tothis extreme some oviparous species lay eggs with late-stage embryos For example the North American greensnake (Liochlorophis vernalis) the New Guinean skink(Sphenomorphus fragilis) and the Australian skink(Saiphos equalis) oviposit when embryos are within weeksor even days of hatching (Sexton and Claypool 1978Guillette 1992 Smith and Shine 1997) At the otherextreme oviposition when embryos are at the gastrulastage occurs in Chamaeleo chamaeleon (Bons and Bons1960) and at comparably early stages for varanid lizards(John Phillips Zoological Society of San Diego unpub-

lished data) A teiid Cnemidophorus uniparens ovipositsat stages 21ndash22 (Billy 1988) and several other species ofCnemidophorus oviposit considerably earlier (NormaManriacutequez Universidad Nacional Autoacutenoma de Meacutexicounpublished data) Reports of what appear to be extraor-dinarily long incubation periods after oviposition suggestthat the eggs of some chameleons may parallel those ofturtles not only in the stage at oviposition (ie gastrula-tion) but also in the potential for embryos to remain in anarrested state even after oviposition (Bons and Bons 1960Ewert 1985)

Despite the wide range of stages over which squamatesoviposit oviposition occurs much more frequently atsome stages than at others (Figure 3) Stages at ovipositiongroup tightly around a mode of stage 30 and the majori-ty of oviparous squamate taxa oviposit at stages 26ndash33This distribution suggests that the stage at oviposition isconstrained One possible constraint is on how early eggscan be laid Perhaps oviposition does not occur muchbefore stage 30 because of the time required to deposit theoviductal proteins (albumen) and the various layers ofshell that surround the ovum In chickens for exampleoviposition occurs 22 hours after ovulation by which timethe embryo has reached the gastrula stage Because reptilesare ectothermic their eggs are likely to require a longerperiod in the oviduct than those of birdsmdashand theirembryos will have attained more advanced embryonicstagesmdashbefore the shell is laid down

For example female Anolis carolinensis and Anolis lim-ifrons lay a sequence of one-egg clutches at minimumintervals of 6ndash7 days (Andrews 1979 1985) and oviposit atstage 30 (Robin M Andrews unpublished data) Ova arereleased into the right and left oviduct alternately suchthat females never have more than one egg per oviductand few females have an egg in both oviducts Thus eggshave a transit time of approximately 6 days (assuming thatno females have two oviductal eggs) to 9 days (assumingthat 50 of the females have two oviductal eggs simulta-neously) Similarly in the laboratory at a body tempera-ture of 30 degC Sceloporus woodi females secrete the pro-teinaceous fibers of the eggshell within the first 24 hoursafter ovulation calcium deposition is maximal on days3ndash9 but continues up to the time of oviposition whenembryos are at stage 27 and have been in the oviduct12ndash14 days (DeMarco 1993 Palmer et al 1993) These andother data (Miller 1985) on sea turtles suggest that reptileeggs have a necessary residence time in the oviduct of 1ndash2weeks before the albumen and shell are deposited

The critical question with respect to what causes ovipo-sition to occur in the vicinity of stage 30 is does oviposi-tion by squamates occur as soon as the shell is completeOr are eggs retained longer Oviposition at stage 30 corre-sponds to retention in the oviduct for 25 (DeMarco1993) to approximately 40 (Shine 1983) of total devel-opmental time (time in utero plus time in nest) Thesepercentages correspond to an egg retention time of 2ndash3

230 BioScience March 2000 Vol 50 No 3

Articles

weeks for small lizards that oviposit at stage 30 and thathave a total developmental time of 50ndash60 days Within aperiod of 2ndash3 weeks embryos will have reached stage 30during a period sufficient to completely shell the eggMany squamates that oviposit at stage 30 however havetotal developmental times that are longer than 50ndash60 days(Shine 1983 DeMarco 1993) If oviposition occurs at25ndash40 of the total developmental period eggs of suchspecies would be in the oviduct for considerably longerthan the 1ndash2 weeks required to completely shell the egg

This line of reasoning suggests that in species thatoviposit earlier than stage 30 either oviposition occurs assoon as eggs are shelled or embryonic development wasarrested before oviposition That is if development fromovulation to oviposition continued normally in theoviduct then after 2 weeks or more well-developedembryos (ie at approximately stage 30) should be pre-sent rather than the gastrula-stage embryos typical ofvaranids and some chameleons For example the periodbetween ovulation and oviposition for varanid lizards isconsistently approximately 1 month (Phillips and Millar1998) despite oviposition at very early stages of develop-ment By contrast squamates that oviposit at stage 30 orlater probably retain eggs for longer than the minimumperiod needed for deposition of the shell Designation ofstage 30 as the benchmark for judging how early eggs arelaid relative to shell formation is of course a crudeapproximation features of the shell or the body tempera-ture of the female could alter the length of time requiredto shell eggs

Another possible constraint that could cause oviposi-tion to cluster around stage 30 is how late in developmenteggs can be laid Such a constraint is particularly impor-tant because it directly limits the evolution of viviparityOnly a few oviparous species retain eggs well into the sec-ond half of development (Figure 3) which suggests thatthe shift from oviparity to viviparity does not occur oftenis rapid or both (Blackburn 1995) In any case the pauci-ty of species that oviposit at stages 34ndash39 suggests thatsuch a strategy presents problems for embryos the repro-ductive female or both

Embryos of oviparous species may experience a numberof problems if they are retained past stages 32ndash33mdasheventhose of species that normally oviposit at these stages

Large absolute increases in the mass of embryos in the sec-ond half of development are associated with greatlyenhanced demands for gas exchange and water (Black etal 1984 Birchard et al 1995) If these demands are notmet development of retained embryos may be retarded(Andrews and Rose 1994) Further retention of eggs at thispoint would be beneficial only if it ldquobought timerdquo for thefemale to locate a suitable oviposition site Such a physio-logical constraint may in addition to having specific con-sequences on embryonic development (see below) help toexplain why so few species lay eggs with embryos beyondstages 32ndash33

Females may also incur problems from extended eggretention The burden of the clutch would not changeappreciably during the first half of development but itwould during the second half when embryos undergolarge absolute increases in mass (Figure 2) as water is tak-en up by the extra-embryonic compartments of eggs and

March 2000 Vol 50 No 3 BioScience 231

Articles

Figure 3 Embryonic stage at oviposition for lizards (top)Lizards that lay a single clutch per reproductive season

(middle) Lizards that lay two or more clutches perreproductive season (bottom) All lizards The figures do notinclude varanids or the three species of Cnemidophorus for

which oviposition occurs before any visible development of theembryo (see text for details) The data are from a literature

review by Shine (1983) and an update to this data set byBlackburn (1995) with more recent data on stage at

oviposition and clutch frequency from Robin M Andrews(unpublished data)

the tissues of the embryo Although retained eggs may takeup less water than oviposited eggs (Mathies and Andrews1996) a gravid female that retains eggs into the secondhalf of development is faced not only with the problem oftransporting a clutch that is increasing in mass but alsowith the potential problem of passing enlarged eggsthrough the pelvic girdle during oviposition (Sinervo andLicht 1991) The rarity of oviposition after stage 34 maytherefore be a direct consequence of increasing egg size

Our concern in this article is largely with the proximatephysiological factors that constrain the embryonic stagesat which eggs are oviposited However it is also importantto recognize that many ultimate factors affect the stage atoviposition For example extended egg retention mayhave a high survival cost for some species because of theburden that the clutch imposes on the female (Sinervo andAdolph 1991) Such a cost would favor oviposition as soonas the shell is formed (as long as suitable nest sites areavailable) Clutch frequency should also affect the stage atoviposition Species that produce multiple clutches perseason could minimize the interval between clutches bylaying eggs at early stages of development thus producingmore clutches within a breeding season On the otherhand for species that are limited to one clutch per seasonthe length of time that eggs spend in the oviduct wouldnot reduce the reproductive output of the female Ifembryos continue to develop in the oviduct in suchspecies then delayed oviposition would shorten the lengthof time that the eggs are in the nest and vulnerable topredators or other mortality agents Single-clutchedspecies would thus be more likely than multiple-clutchedspecies to oviposit when embryos are at advanced stages

This prediction is borne out by data on lizards (Figure3) few multiple-clutched lizard species oviposit later thanstage 30 whereas most single-clutched lizard speciesoviposit at stage 30 or later (X2 = 70 P lt 001 2 acute 2 con-tingency analysis two-tailed test) Despite the associationbetween clutch frequency and stage at oviposition webelieve that selection for oviposition to occur at very earlyor very late stages of embryonic development will seldombe effective because of the time necessary to complete theegg shell on the one hand and physiological constraintson embryonic development in utero on the other

Consequences of extended egg retentionfor sceloporine lizardsOur research focuses on spiny lizards of the genus Scelo-porus and tree lizards of the genus Urosaurus Sceloporusand Urosaurus are sister taxa that is each otherrsquos closestrelative Urosaurus contains approximately 10 species allof which are oviparous Sceloporus contains approximate-ly 70 species almost half of which are viviparous (Sites etal 1992) Viviparity has apparently arisen only four timesin this genus once in the ancestor of the formosus grouponce in the ancestor of the grammicus megalepidurus andtorquatus groups and twice within the scalaris group

(Meacutendez-de la Cruz et al 1998) The scalaris group is theonly one to have both oviparous and viviparous speciesand the close relationship of the oviparous and viviparousspecies suggests that viviparity is of recent origin in thisgroup (Benabib et al 1997) The few origins of viviparityand the large number of species groups (approximately15) in which all members are oviparous suggests that thetransition to viviparity may be difficult in Sceloporus

Because oviposition at successively later stages ofembryonic development is the putative transition toviviparity the present-day consequences of extended eggretention on embryonic development provide insightsinto how easy it is for viviparity to evolve We have there-fore studied the consequences of the retention of eggsbeyond the normal stage of oviposition We initiallyfocused on oviparous species known to retain eggs toadvanced stages but a more complete picture requiredthat we also look at the consequences of egg retention bythe more typical speciesmdashthat is those that oviposit whenembryos reach approximately stage 30

The experimental protocol common to all of our stud-ies has been to keep gravid females of each species in thelaboratory under conditions that inhibit oviposition Thisapproach allowed us to assess the capacity of females tosupport embryonic development in utero past the time ofnormal oviposition To inhibit oviposition we tookadvantage of a normal response to drought because suc-cessful nesting in the field is dependent on adequate soilmoisture and rainfall is often unpredictable femalesrespond to drought by retaining their eggs (Andrews andRose 1994) Experimental females were thus kept in cageswith dry soil as a substrate which simulated drought con-ditions and inhibited oviposition (Andrews and Rose1994) Control females were housed under the same con-ditions as experimental females except that the cage sub-strate was moist soil most females oviposited normallyunder these conditions Eggs from control females wereincubated in a standard incubation medium and at thesame temperature as eggs within experimental females

One set of experiments examined egg retention bySceloporus scalaris an oviparous species that is a model forthe transition from oviparity to viviparity (Figure 4) Sscalaris a member of the scalaris species group occurs ingrasslands at many elevations in Arizona New Mexicoand northern Mexico Females from a low-elevation pop-ulation are able to retain eggs for an unprecedented lengthof time although they normally oviposit at stages 31ndash33(Mathies and Andrews 1995 1996 Thomas Mathies andRobin M Andrews unpublished data) Eggs retained aslong as 1 month (to stage 38) and surgically removed fromfemales develop at the same rate and have the same sur-vival to hatching rate as control eggs Eggs retained for afurther 10ndash13 days advance to stage 40 these eggs pro-duced viable hatchlings 1ndash3 days after they were surgicallyremoved from females When the same experiment wasconducted at a slightly higher incubation temperature

232 BioScience March 2000 Vol 50 No 3

Articles

development was somewhat slowerfor retained embryos than for con-trol embryos presumably becausethe higher incubation temperatureincreased the oxygen requirementsof embryos beyond that availablein utero (Andrews 1997)

Nevertheless among squamatesS scalaris has an unusual ability toextend egg retention to within afew days of hatching and to do sowith little or no impairment ofembryonic development The abil-ity to support embryogenesis inutero to advanced stages is alsofound in another oviparous speciesin the scalaris group Sceloporusaeneus (Robin M Andrewsunpublished observation) the sis-ter species of the viviparous Scelo-porus bicanthalis The ability ofoviparous species in the scalarisgroup to extend egg retention toadvanced embryonic stages has likely facilitated the evolu-tion of viviparity in this group

Sceloporus virgatus an oviparous member of the undu-latus species group also normally lays eggs at stages 31ndash33In contrast to S scalaris however when eggs of S virgatusare retained past the normal time of oviposition as long asto stage 37 embryonic development is retarded (Figure 4Andrews and Rose 1994 Andrews 1997) For example thedry masses of embryos retained for 30 days past the nor-mal time of oviposition were less than half those of con-trols and retained embryos showed considerably lowersurvival to hatching than control eggs Female S virgatusthat retain eggs beyond the normal time of ovipositionthus have greatly reduced fitness This assertion is not toimply that extended egg retention has no benefit in thisspeciesmdashfemales that do not retain eggs until the seasonalrains occur would have zero fitness for that reproductiveseason

S virgatus appears to be unusual both within the undu-latus species group and within the genus Sceloporus in itsability to normally support embryonic development paststage 30 (Mathies 1998) No other species that we studiedin the undulatus species group is able to retain eggs appre-ciably beyond stage 30 For example females of Sceloporusoccidentalis another member of the undulatus groupoviposit when embryos reach stages 28ndash30 even if condi-tions are not suitable for nesting This pattern is also char-acteristic of Sceloporus graciosus (a member of the gracio-sus species group) Females of two members of theundulatus species groupmdashSceloporus undulatus consobri-nus from Arizona and Sceloporus undulatus hyacinthinusfrom Virginia and Missourimdashboth typically oviposit whenembryos are at stage 28ndash30 (Sexton and Marion 1974

Mathies 1998) as do female Sceloporus clarki (a member ofthe clarki species group) Moreover embryos of thesespecies do not develop further when eggs are retainedbeyond the normal time of oviposition (Mathies 1998)Initially the similar stages at oviposition for S u consobri-nus and S u hyacinthinus housed on moist and dry sub-strates would appear to be the result of the brief periodapproximately 10 days of facultative egg retention underdrought conditions However during this 10-day periodembryos in control eggs had reached stage 32 and had drymasses four to eight times greater than those of theretained embryos These observations indicate that devel-opment of retained embryos was arrested Females in theundulatus graciosus and clarki species groups thus exhib-it a limited ability to extend egg retention and to supportembryogenesis in utero beyond stage 30

In contrast to members of the undulatus group a mem-ber of the sister genus to Sceloporus Urosaurus ornatus isable to retain eggs for at least a month past the time ofnormal oviposition (Mathies and Andrews 1999) As inmost members of the undulatus group however develop-ment is arrested at approximately the time of normal ovipo-sition In this case development is arrested at stage 305whereas control embryos reach stage 38 after the samelength of incubation (Figure 4) When retained eggs arelaid development is reinitiated and continues to hatchingHatching success of retained eggs is as high as that of con-trol eggs Retention in Urosaurus thus has no obvious effecton embryonic development other than to delay hatchingand then only by the time spent in arrest Retention doesaffect hatchlings however Hatchlings produced after longperiods of retention are smaller and less hydrated thanhatchlings produced after short periods of retention

March 2000 Vol 50 No 3 BioScience 233

Articles

Figure 4 Embryo stage in three species of phrynosomatid lizards as a function of dayselapsed since the normal time of oviposition Eggs of control females (open circles)were oviposited at the normal time Eggs of experimental females (solid circles) wereretained in the oviducts for the number of days indicated Oviposition byexperimental females was inhibited by providing dry substrates for nesting Thedevelopment of the retained embryos was arrested for Urosaurus ornatus (left) wasretarded for Sceloporus virgatus (middle) and proceeded normally for Sceloporusscalaris (right) Figure modified from Andrews and Rose (1994) Mathies and Andrews(1996) and Mathies (1998)

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

turtle Chelodina rugosa in Australia provide direct evi-dence that hypoxia maintains developmental arrest in thisspecies (Kennett et al 1993) The rigid-shelled eggs arelaid underwater in seasonal ponds and the embryosremain in developmental arrest as long as the eggs areimmersed in water Embryos resume development onlyafter the soil dries and oxygen tension rises In an experi-ment in which eggs were shifted from water to an atmos-phere of pure nitrogen embryos remained in develop-mental arrest they resumed development only whenexposed to atmospheric air

An alternative or perhaps complementary explanationfor oviposition early in development by turtles and croco-dilians concerns the adhesion of the embryo and thevitelline membrane to the shell (Ewert 1985 Ferguson1985) Adhesion shortly after oviposition appears to havea respiratory function because the ldquochalkingrdquo (drying) ofthe shell that is associated with adhesion increases its con-ductance to gases (Thompson 1985) Because movementof an egg after oviposition prevents adhesion retention ofeggs in the oviduct where they are inevitably jostledwould probably preclude adhesion and further develop-ment This may be a moot point however because oviduc-tal fluids would also prevent chalking as long as the eggremained in the oviduct

Another potential constraint on extended egg retentionfor crocodilians and turtles is that much of the calciumused in development is mobilized from the shell (Packardand Packard 1984) If calcification of the shell of turtle andcrocodilian embryos were reduced as a means to enhancegas exchange in the oviduct then the later growth of theembryo would be limited Calcium does not limit earlydevelopment however and selection could presumablyincrease the amount of calcium stored in the yolk Never-theless the suite of physiological and morphological fea-tures in turtles crocodilians and sphenodontidans mayform a substantial barrier to appreciable embryonic devel-opment in the oviducts for these reptiles

Squamates Oviposition at early developmental stagesmay explain why no crocodilian turtle or sphenodonti-dan has evolved viviparity (Shine 1983) On the otherhand squamate reptiles exhibit nearly the entire gamut ofpossible embryonic stages at oviposition At one extremeapproximately 20 of squamates are viviparous Close tothis extreme some oviparous species lay eggs with late-stage embryos For example the North American greensnake (Liochlorophis vernalis) the New Guinean skink(Sphenomorphus fragilis) and the Australian skink(Saiphos equalis) oviposit when embryos are within weeksor even days of hatching (Sexton and Claypool 1978Guillette 1992 Smith and Shine 1997) At the otherextreme oviposition when embryos are at the gastrulastage occurs in Chamaeleo chamaeleon (Bons and Bons1960) and at comparably early stages for varanid lizards(John Phillips Zoological Society of San Diego unpub-

lished data) A teiid Cnemidophorus uniparens ovipositsat stages 21ndash22 (Billy 1988) and several other species ofCnemidophorus oviposit considerably earlier (NormaManriacutequez Universidad Nacional Autoacutenoma de Meacutexicounpublished data) Reports of what appear to be extraor-dinarily long incubation periods after oviposition suggestthat the eggs of some chameleons may parallel those ofturtles not only in the stage at oviposition (ie gastrula-tion) but also in the potential for embryos to remain in anarrested state even after oviposition (Bons and Bons 1960Ewert 1985)

Despite the wide range of stages over which squamatesoviposit oviposition occurs much more frequently atsome stages than at others (Figure 3) Stages at ovipositiongroup tightly around a mode of stage 30 and the majori-ty of oviparous squamate taxa oviposit at stages 26ndash33This distribution suggests that the stage at oviposition isconstrained One possible constraint is on how early eggscan be laid Perhaps oviposition does not occur muchbefore stage 30 because of the time required to deposit theoviductal proteins (albumen) and the various layers ofshell that surround the ovum In chickens for exampleoviposition occurs 22 hours after ovulation by which timethe embryo has reached the gastrula stage Because reptilesare ectothermic their eggs are likely to require a longerperiod in the oviduct than those of birdsmdashand theirembryos will have attained more advanced embryonicstagesmdashbefore the shell is laid down

For example female Anolis carolinensis and Anolis lim-ifrons lay a sequence of one-egg clutches at minimumintervals of 6ndash7 days (Andrews 1979 1985) and oviposit atstage 30 (Robin M Andrews unpublished data) Ova arereleased into the right and left oviduct alternately suchthat females never have more than one egg per oviductand few females have an egg in both oviducts Thus eggshave a transit time of approximately 6 days (assuming thatno females have two oviductal eggs) to 9 days (assumingthat 50 of the females have two oviductal eggs simulta-neously) Similarly in the laboratory at a body tempera-ture of 30 degC Sceloporus woodi females secrete the pro-teinaceous fibers of the eggshell within the first 24 hoursafter ovulation calcium deposition is maximal on days3ndash9 but continues up to the time of oviposition whenembryos are at stage 27 and have been in the oviduct12ndash14 days (DeMarco 1993 Palmer et al 1993) These andother data (Miller 1985) on sea turtles suggest that reptileeggs have a necessary residence time in the oviduct of 1ndash2weeks before the albumen and shell are deposited

The critical question with respect to what causes ovipo-sition to occur in the vicinity of stage 30 is does oviposi-tion by squamates occur as soon as the shell is completeOr are eggs retained longer Oviposition at stage 30 corre-sponds to retention in the oviduct for 25 (DeMarco1993) to approximately 40 (Shine 1983) of total devel-opmental time (time in utero plus time in nest) Thesepercentages correspond to an egg retention time of 2ndash3

230 BioScience March 2000 Vol 50 No 3

Articles

weeks for small lizards that oviposit at stage 30 and thathave a total developmental time of 50ndash60 days Within aperiod of 2ndash3 weeks embryos will have reached stage 30during a period sufficient to completely shell the eggMany squamates that oviposit at stage 30 however havetotal developmental times that are longer than 50ndash60 days(Shine 1983 DeMarco 1993) If oviposition occurs at25ndash40 of the total developmental period eggs of suchspecies would be in the oviduct for considerably longerthan the 1ndash2 weeks required to completely shell the egg

This line of reasoning suggests that in species thatoviposit earlier than stage 30 either oviposition occurs assoon as eggs are shelled or embryonic development wasarrested before oviposition That is if development fromovulation to oviposition continued normally in theoviduct then after 2 weeks or more well-developedembryos (ie at approximately stage 30) should be pre-sent rather than the gastrula-stage embryos typical ofvaranids and some chameleons For example the periodbetween ovulation and oviposition for varanid lizards isconsistently approximately 1 month (Phillips and Millar1998) despite oviposition at very early stages of develop-ment By contrast squamates that oviposit at stage 30 orlater probably retain eggs for longer than the minimumperiod needed for deposition of the shell Designation ofstage 30 as the benchmark for judging how early eggs arelaid relative to shell formation is of course a crudeapproximation features of the shell or the body tempera-ture of the female could alter the length of time requiredto shell eggs

Another possible constraint that could cause oviposi-tion to cluster around stage 30 is how late in developmenteggs can be laid Such a constraint is particularly impor-tant because it directly limits the evolution of viviparityOnly a few oviparous species retain eggs well into the sec-ond half of development (Figure 3) which suggests thatthe shift from oviparity to viviparity does not occur oftenis rapid or both (Blackburn 1995) In any case the pauci-ty of species that oviposit at stages 34ndash39 suggests thatsuch a strategy presents problems for embryos the repro-ductive female or both

Embryos of oviparous species may experience a numberof problems if they are retained past stages 32ndash33mdasheventhose of species that normally oviposit at these stages

Large absolute increases in the mass of embryos in the sec-ond half of development are associated with greatlyenhanced demands for gas exchange and water (Black etal 1984 Birchard et al 1995) If these demands are notmet development of retained embryos may be retarded(Andrews and Rose 1994) Further retention of eggs at thispoint would be beneficial only if it ldquobought timerdquo for thefemale to locate a suitable oviposition site Such a physio-logical constraint may in addition to having specific con-sequences on embryonic development (see below) help toexplain why so few species lay eggs with embryos beyondstages 32ndash33

Females may also incur problems from extended eggretention The burden of the clutch would not changeappreciably during the first half of development but itwould during the second half when embryos undergolarge absolute increases in mass (Figure 2) as water is tak-en up by the extra-embryonic compartments of eggs and

March 2000 Vol 50 No 3 BioScience 231

Articles

Figure 3 Embryonic stage at oviposition for lizards (top)Lizards that lay a single clutch per reproductive season

(middle) Lizards that lay two or more clutches perreproductive season (bottom) All lizards The figures do notinclude varanids or the three species of Cnemidophorus for

which oviposition occurs before any visible development of theembryo (see text for details) The data are from a literature

review by Shine (1983) and an update to this data set byBlackburn (1995) with more recent data on stage at

oviposition and clutch frequency from Robin M Andrews(unpublished data)

the tissues of the embryo Although retained eggs may takeup less water than oviposited eggs (Mathies and Andrews1996) a gravid female that retains eggs into the secondhalf of development is faced not only with the problem oftransporting a clutch that is increasing in mass but alsowith the potential problem of passing enlarged eggsthrough the pelvic girdle during oviposition (Sinervo andLicht 1991) The rarity of oviposition after stage 34 maytherefore be a direct consequence of increasing egg size

Our concern in this article is largely with the proximatephysiological factors that constrain the embryonic stagesat which eggs are oviposited However it is also importantto recognize that many ultimate factors affect the stage atoviposition For example extended egg retention mayhave a high survival cost for some species because of theburden that the clutch imposes on the female (Sinervo andAdolph 1991) Such a cost would favor oviposition as soonas the shell is formed (as long as suitable nest sites areavailable) Clutch frequency should also affect the stage atoviposition Species that produce multiple clutches perseason could minimize the interval between clutches bylaying eggs at early stages of development thus producingmore clutches within a breeding season On the otherhand for species that are limited to one clutch per seasonthe length of time that eggs spend in the oviduct wouldnot reduce the reproductive output of the female Ifembryos continue to develop in the oviduct in suchspecies then delayed oviposition would shorten the lengthof time that the eggs are in the nest and vulnerable topredators or other mortality agents Single-clutchedspecies would thus be more likely than multiple-clutchedspecies to oviposit when embryos are at advanced stages

This prediction is borne out by data on lizards (Figure3) few multiple-clutched lizard species oviposit later thanstage 30 whereas most single-clutched lizard speciesoviposit at stage 30 or later (X2 = 70 P lt 001 2 acute 2 con-tingency analysis two-tailed test) Despite the associationbetween clutch frequency and stage at oviposition webelieve that selection for oviposition to occur at very earlyor very late stages of embryonic development will seldombe effective because of the time necessary to complete theegg shell on the one hand and physiological constraintson embryonic development in utero on the other

Consequences of extended egg retentionfor sceloporine lizardsOur research focuses on spiny lizards of the genus Scelo-porus and tree lizards of the genus Urosaurus Sceloporusand Urosaurus are sister taxa that is each otherrsquos closestrelative Urosaurus contains approximately 10 species allof which are oviparous Sceloporus contains approximate-ly 70 species almost half of which are viviparous (Sites etal 1992) Viviparity has apparently arisen only four timesin this genus once in the ancestor of the formosus grouponce in the ancestor of the grammicus megalepidurus andtorquatus groups and twice within the scalaris group

(Meacutendez-de la Cruz et al 1998) The scalaris group is theonly one to have both oviparous and viviparous speciesand the close relationship of the oviparous and viviparousspecies suggests that viviparity is of recent origin in thisgroup (Benabib et al 1997) The few origins of viviparityand the large number of species groups (approximately15) in which all members are oviparous suggests that thetransition to viviparity may be difficult in Sceloporus

Because oviposition at successively later stages ofembryonic development is the putative transition toviviparity the present-day consequences of extended eggretention on embryonic development provide insightsinto how easy it is for viviparity to evolve We have there-fore studied the consequences of the retention of eggsbeyond the normal stage of oviposition We initiallyfocused on oviparous species known to retain eggs toadvanced stages but a more complete picture requiredthat we also look at the consequences of egg retention bythe more typical speciesmdashthat is those that oviposit whenembryos reach approximately stage 30

The experimental protocol common to all of our stud-ies has been to keep gravid females of each species in thelaboratory under conditions that inhibit oviposition Thisapproach allowed us to assess the capacity of females tosupport embryonic development in utero past the time ofnormal oviposition To inhibit oviposition we tookadvantage of a normal response to drought because suc-cessful nesting in the field is dependent on adequate soilmoisture and rainfall is often unpredictable femalesrespond to drought by retaining their eggs (Andrews andRose 1994) Experimental females were thus kept in cageswith dry soil as a substrate which simulated drought con-ditions and inhibited oviposition (Andrews and Rose1994) Control females were housed under the same con-ditions as experimental females except that the cage sub-strate was moist soil most females oviposited normallyunder these conditions Eggs from control females wereincubated in a standard incubation medium and at thesame temperature as eggs within experimental females

One set of experiments examined egg retention bySceloporus scalaris an oviparous species that is a model forthe transition from oviparity to viviparity (Figure 4) Sscalaris a member of the scalaris species group occurs ingrasslands at many elevations in Arizona New Mexicoand northern Mexico Females from a low-elevation pop-ulation are able to retain eggs for an unprecedented lengthof time although they normally oviposit at stages 31ndash33(Mathies and Andrews 1995 1996 Thomas Mathies andRobin M Andrews unpublished data) Eggs retained aslong as 1 month (to stage 38) and surgically removed fromfemales develop at the same rate and have the same sur-vival to hatching rate as control eggs Eggs retained for afurther 10ndash13 days advance to stage 40 these eggs pro-duced viable hatchlings 1ndash3 days after they were surgicallyremoved from females When the same experiment wasconducted at a slightly higher incubation temperature

232 BioScience March 2000 Vol 50 No 3

Articles

development was somewhat slowerfor retained embryos than for con-trol embryos presumably becausethe higher incubation temperatureincreased the oxygen requirementsof embryos beyond that availablein utero (Andrews 1997)

Nevertheless among squamatesS scalaris has an unusual ability toextend egg retention to within afew days of hatching and to do sowith little or no impairment ofembryonic development The abil-ity to support embryogenesis inutero to advanced stages is alsofound in another oviparous speciesin the scalaris group Sceloporusaeneus (Robin M Andrewsunpublished observation) the sis-ter species of the viviparous Scelo-porus bicanthalis The ability ofoviparous species in the scalarisgroup to extend egg retention toadvanced embryonic stages has likely facilitated the evolu-tion of viviparity in this group

Sceloporus virgatus an oviparous member of the undu-latus species group also normally lays eggs at stages 31ndash33In contrast to S scalaris however when eggs of S virgatusare retained past the normal time of oviposition as long asto stage 37 embryonic development is retarded (Figure 4Andrews and Rose 1994 Andrews 1997) For example thedry masses of embryos retained for 30 days past the nor-mal time of oviposition were less than half those of con-trols and retained embryos showed considerably lowersurvival to hatching than control eggs Female S virgatusthat retain eggs beyond the normal time of ovipositionthus have greatly reduced fitness This assertion is not toimply that extended egg retention has no benefit in thisspeciesmdashfemales that do not retain eggs until the seasonalrains occur would have zero fitness for that reproductiveseason

S virgatus appears to be unusual both within the undu-latus species group and within the genus Sceloporus in itsability to normally support embryonic development paststage 30 (Mathies 1998) No other species that we studiedin the undulatus species group is able to retain eggs appre-ciably beyond stage 30 For example females of Sceloporusoccidentalis another member of the undulatus groupoviposit when embryos reach stages 28ndash30 even if condi-tions are not suitable for nesting This pattern is also char-acteristic of Sceloporus graciosus (a member of the gracio-sus species group) Females of two members of theundulatus species groupmdashSceloporus undulatus consobri-nus from Arizona and Sceloporus undulatus hyacinthinusfrom Virginia and Missourimdashboth typically oviposit whenembryos are at stage 28ndash30 (Sexton and Marion 1974

Mathies 1998) as do female Sceloporus clarki (a member ofthe clarki species group) Moreover embryos of thesespecies do not develop further when eggs are retainedbeyond the normal time of oviposition (Mathies 1998)Initially the similar stages at oviposition for S u consobri-nus and S u hyacinthinus housed on moist and dry sub-strates would appear to be the result of the brief periodapproximately 10 days of facultative egg retention underdrought conditions However during this 10-day periodembryos in control eggs had reached stage 32 and had drymasses four to eight times greater than those of theretained embryos These observations indicate that devel-opment of retained embryos was arrested Females in theundulatus graciosus and clarki species groups thus exhib-it a limited ability to extend egg retention and to supportembryogenesis in utero beyond stage 30

In contrast to members of the undulatus group a mem-ber of the sister genus to Sceloporus Urosaurus ornatus isable to retain eggs for at least a month past the time ofnormal oviposition (Mathies and Andrews 1999) As inmost members of the undulatus group however develop-ment is arrested at approximately the time of normal ovipo-sition In this case development is arrested at stage 305whereas control embryos reach stage 38 after the samelength of incubation (Figure 4) When retained eggs arelaid development is reinitiated and continues to hatchingHatching success of retained eggs is as high as that of con-trol eggs Retention in Urosaurus thus has no obvious effecton embryonic development other than to delay hatchingand then only by the time spent in arrest Retention doesaffect hatchlings however Hatchlings produced after longperiods of retention are smaller and less hydrated thanhatchlings produced after short periods of retention

March 2000 Vol 50 No 3 BioScience 233

Articles

Figure 4 Embryo stage in three species of phrynosomatid lizards as a function of dayselapsed since the normal time of oviposition Eggs of control females (open circles)were oviposited at the normal time Eggs of experimental females (solid circles) wereretained in the oviducts for the number of days indicated Oviposition byexperimental females was inhibited by providing dry substrates for nesting Thedevelopment of the retained embryos was arrested for Urosaurus ornatus (left) wasretarded for Sceloporus virgatus (middle) and proceeded normally for Sceloporusscalaris (right) Figure modified from Andrews and Rose (1994) Mathies and Andrews(1996) and Mathies (1998)

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

weeks for small lizards that oviposit at stage 30 and thathave a total developmental time of 50ndash60 days Within aperiod of 2ndash3 weeks embryos will have reached stage 30during a period sufficient to completely shell the eggMany squamates that oviposit at stage 30 however havetotal developmental times that are longer than 50ndash60 days(Shine 1983 DeMarco 1993) If oviposition occurs at25ndash40 of the total developmental period eggs of suchspecies would be in the oviduct for considerably longerthan the 1ndash2 weeks required to completely shell the egg

This line of reasoning suggests that in species thatoviposit earlier than stage 30 either oviposition occurs assoon as eggs are shelled or embryonic development wasarrested before oviposition That is if development fromovulation to oviposition continued normally in theoviduct then after 2 weeks or more well-developedembryos (ie at approximately stage 30) should be pre-sent rather than the gastrula-stage embryos typical ofvaranids and some chameleons For example the periodbetween ovulation and oviposition for varanid lizards isconsistently approximately 1 month (Phillips and Millar1998) despite oviposition at very early stages of develop-ment By contrast squamates that oviposit at stage 30 orlater probably retain eggs for longer than the minimumperiod needed for deposition of the shell Designation ofstage 30 as the benchmark for judging how early eggs arelaid relative to shell formation is of course a crudeapproximation features of the shell or the body tempera-ture of the female could alter the length of time requiredto shell eggs

Another possible constraint that could cause oviposi-tion to cluster around stage 30 is how late in developmenteggs can be laid Such a constraint is particularly impor-tant because it directly limits the evolution of viviparityOnly a few oviparous species retain eggs well into the sec-ond half of development (Figure 3) which suggests thatthe shift from oviparity to viviparity does not occur oftenis rapid or both (Blackburn 1995) In any case the pauci-ty of species that oviposit at stages 34ndash39 suggests thatsuch a strategy presents problems for embryos the repro-ductive female or both

Embryos of oviparous species may experience a numberof problems if they are retained past stages 32ndash33mdasheventhose of species that normally oviposit at these stages

Large absolute increases in the mass of embryos in the sec-ond half of development are associated with greatlyenhanced demands for gas exchange and water (Black etal 1984 Birchard et al 1995) If these demands are notmet development of retained embryos may be retarded(Andrews and Rose 1994) Further retention of eggs at thispoint would be beneficial only if it ldquobought timerdquo for thefemale to locate a suitable oviposition site Such a physio-logical constraint may in addition to having specific con-sequences on embryonic development (see below) help toexplain why so few species lay eggs with embryos beyondstages 32ndash33

Females may also incur problems from extended eggretention The burden of the clutch would not changeappreciably during the first half of development but itwould during the second half when embryos undergolarge absolute increases in mass (Figure 2) as water is tak-en up by the extra-embryonic compartments of eggs and

March 2000 Vol 50 No 3 BioScience 231

Articles

Figure 3 Embryonic stage at oviposition for lizards (top)Lizards that lay a single clutch per reproductive season

(middle) Lizards that lay two or more clutches perreproductive season (bottom) All lizards The figures do notinclude varanids or the three species of Cnemidophorus for

which oviposition occurs before any visible development of theembryo (see text for details) The data are from a literature

review by Shine (1983) and an update to this data set byBlackburn (1995) with more recent data on stage at

oviposition and clutch frequency from Robin M Andrews(unpublished data)

the tissues of the embryo Although retained eggs may takeup less water than oviposited eggs (Mathies and Andrews1996) a gravid female that retains eggs into the secondhalf of development is faced not only with the problem oftransporting a clutch that is increasing in mass but alsowith the potential problem of passing enlarged eggsthrough the pelvic girdle during oviposition (Sinervo andLicht 1991) The rarity of oviposition after stage 34 maytherefore be a direct consequence of increasing egg size

Our concern in this article is largely with the proximatephysiological factors that constrain the embryonic stagesat which eggs are oviposited However it is also importantto recognize that many ultimate factors affect the stage atoviposition For example extended egg retention mayhave a high survival cost for some species because of theburden that the clutch imposes on the female (Sinervo andAdolph 1991) Such a cost would favor oviposition as soonas the shell is formed (as long as suitable nest sites areavailable) Clutch frequency should also affect the stage atoviposition Species that produce multiple clutches perseason could minimize the interval between clutches bylaying eggs at early stages of development thus producingmore clutches within a breeding season On the otherhand for species that are limited to one clutch per seasonthe length of time that eggs spend in the oviduct wouldnot reduce the reproductive output of the female Ifembryos continue to develop in the oviduct in suchspecies then delayed oviposition would shorten the lengthof time that the eggs are in the nest and vulnerable topredators or other mortality agents Single-clutchedspecies would thus be more likely than multiple-clutchedspecies to oviposit when embryos are at advanced stages

This prediction is borne out by data on lizards (Figure3) few multiple-clutched lizard species oviposit later thanstage 30 whereas most single-clutched lizard speciesoviposit at stage 30 or later (X2 = 70 P lt 001 2 acute 2 con-tingency analysis two-tailed test) Despite the associationbetween clutch frequency and stage at oviposition webelieve that selection for oviposition to occur at very earlyor very late stages of embryonic development will seldombe effective because of the time necessary to complete theegg shell on the one hand and physiological constraintson embryonic development in utero on the other

Consequences of extended egg retentionfor sceloporine lizardsOur research focuses on spiny lizards of the genus Scelo-porus and tree lizards of the genus Urosaurus Sceloporusand Urosaurus are sister taxa that is each otherrsquos closestrelative Urosaurus contains approximately 10 species allof which are oviparous Sceloporus contains approximate-ly 70 species almost half of which are viviparous (Sites etal 1992) Viviparity has apparently arisen only four timesin this genus once in the ancestor of the formosus grouponce in the ancestor of the grammicus megalepidurus andtorquatus groups and twice within the scalaris group

(Meacutendez-de la Cruz et al 1998) The scalaris group is theonly one to have both oviparous and viviparous speciesand the close relationship of the oviparous and viviparousspecies suggests that viviparity is of recent origin in thisgroup (Benabib et al 1997) The few origins of viviparityand the large number of species groups (approximately15) in which all members are oviparous suggests that thetransition to viviparity may be difficult in Sceloporus

Because oviposition at successively later stages ofembryonic development is the putative transition toviviparity the present-day consequences of extended eggretention on embryonic development provide insightsinto how easy it is for viviparity to evolve We have there-fore studied the consequences of the retention of eggsbeyond the normal stage of oviposition We initiallyfocused on oviparous species known to retain eggs toadvanced stages but a more complete picture requiredthat we also look at the consequences of egg retention bythe more typical speciesmdashthat is those that oviposit whenembryos reach approximately stage 30

The experimental protocol common to all of our stud-ies has been to keep gravid females of each species in thelaboratory under conditions that inhibit oviposition Thisapproach allowed us to assess the capacity of females tosupport embryonic development in utero past the time ofnormal oviposition To inhibit oviposition we tookadvantage of a normal response to drought because suc-cessful nesting in the field is dependent on adequate soilmoisture and rainfall is often unpredictable femalesrespond to drought by retaining their eggs (Andrews andRose 1994) Experimental females were thus kept in cageswith dry soil as a substrate which simulated drought con-ditions and inhibited oviposition (Andrews and Rose1994) Control females were housed under the same con-ditions as experimental females except that the cage sub-strate was moist soil most females oviposited normallyunder these conditions Eggs from control females wereincubated in a standard incubation medium and at thesame temperature as eggs within experimental females

One set of experiments examined egg retention bySceloporus scalaris an oviparous species that is a model forthe transition from oviparity to viviparity (Figure 4) Sscalaris a member of the scalaris species group occurs ingrasslands at many elevations in Arizona New Mexicoand northern Mexico Females from a low-elevation pop-ulation are able to retain eggs for an unprecedented lengthof time although they normally oviposit at stages 31ndash33(Mathies and Andrews 1995 1996 Thomas Mathies andRobin M Andrews unpublished data) Eggs retained aslong as 1 month (to stage 38) and surgically removed fromfemales develop at the same rate and have the same sur-vival to hatching rate as control eggs Eggs retained for afurther 10ndash13 days advance to stage 40 these eggs pro-duced viable hatchlings 1ndash3 days after they were surgicallyremoved from females When the same experiment wasconducted at a slightly higher incubation temperature

232 BioScience March 2000 Vol 50 No 3

Articles

development was somewhat slowerfor retained embryos than for con-trol embryos presumably becausethe higher incubation temperatureincreased the oxygen requirementsof embryos beyond that availablein utero (Andrews 1997)

Nevertheless among squamatesS scalaris has an unusual ability toextend egg retention to within afew days of hatching and to do sowith little or no impairment ofembryonic development The abil-ity to support embryogenesis inutero to advanced stages is alsofound in another oviparous speciesin the scalaris group Sceloporusaeneus (Robin M Andrewsunpublished observation) the sis-ter species of the viviparous Scelo-porus bicanthalis The ability ofoviparous species in the scalarisgroup to extend egg retention toadvanced embryonic stages has likely facilitated the evolu-tion of viviparity in this group

Sceloporus virgatus an oviparous member of the undu-latus species group also normally lays eggs at stages 31ndash33In contrast to S scalaris however when eggs of S virgatusare retained past the normal time of oviposition as long asto stage 37 embryonic development is retarded (Figure 4Andrews and Rose 1994 Andrews 1997) For example thedry masses of embryos retained for 30 days past the nor-mal time of oviposition were less than half those of con-trols and retained embryos showed considerably lowersurvival to hatching than control eggs Female S virgatusthat retain eggs beyond the normal time of ovipositionthus have greatly reduced fitness This assertion is not toimply that extended egg retention has no benefit in thisspeciesmdashfemales that do not retain eggs until the seasonalrains occur would have zero fitness for that reproductiveseason

S virgatus appears to be unusual both within the undu-latus species group and within the genus Sceloporus in itsability to normally support embryonic development paststage 30 (Mathies 1998) No other species that we studiedin the undulatus species group is able to retain eggs appre-ciably beyond stage 30 For example females of Sceloporusoccidentalis another member of the undulatus groupoviposit when embryos reach stages 28ndash30 even if condi-tions are not suitable for nesting This pattern is also char-acteristic of Sceloporus graciosus (a member of the gracio-sus species group) Females of two members of theundulatus species groupmdashSceloporus undulatus consobri-nus from Arizona and Sceloporus undulatus hyacinthinusfrom Virginia and Missourimdashboth typically oviposit whenembryos are at stage 28ndash30 (Sexton and Marion 1974

Mathies 1998) as do female Sceloporus clarki (a member ofthe clarki species group) Moreover embryos of thesespecies do not develop further when eggs are retainedbeyond the normal time of oviposition (Mathies 1998)Initially the similar stages at oviposition for S u consobri-nus and S u hyacinthinus housed on moist and dry sub-strates would appear to be the result of the brief periodapproximately 10 days of facultative egg retention underdrought conditions However during this 10-day periodembryos in control eggs had reached stage 32 and had drymasses four to eight times greater than those of theretained embryos These observations indicate that devel-opment of retained embryos was arrested Females in theundulatus graciosus and clarki species groups thus exhib-it a limited ability to extend egg retention and to supportembryogenesis in utero beyond stage 30

In contrast to members of the undulatus group a mem-ber of the sister genus to Sceloporus Urosaurus ornatus isable to retain eggs for at least a month past the time ofnormal oviposition (Mathies and Andrews 1999) As inmost members of the undulatus group however develop-ment is arrested at approximately the time of normal ovipo-sition In this case development is arrested at stage 305whereas control embryos reach stage 38 after the samelength of incubation (Figure 4) When retained eggs arelaid development is reinitiated and continues to hatchingHatching success of retained eggs is as high as that of con-trol eggs Retention in Urosaurus thus has no obvious effecton embryonic development other than to delay hatchingand then only by the time spent in arrest Retention doesaffect hatchlings however Hatchlings produced after longperiods of retention are smaller and less hydrated thanhatchlings produced after short periods of retention

March 2000 Vol 50 No 3 BioScience 233

Articles

Figure 4 Embryo stage in three species of phrynosomatid lizards as a function of dayselapsed since the normal time of oviposition Eggs of control females (open circles)were oviposited at the normal time Eggs of experimental females (solid circles) wereretained in the oviducts for the number of days indicated Oviposition byexperimental females was inhibited by providing dry substrates for nesting Thedevelopment of the retained embryos was arrested for Urosaurus ornatus (left) wasretarded for Sceloporus virgatus (middle) and proceeded normally for Sceloporusscalaris (right) Figure modified from Andrews and Rose (1994) Mathies and Andrews(1996) and Mathies (1998)

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

the tissues of the embryo Although retained eggs may takeup less water than oviposited eggs (Mathies and Andrews1996) a gravid female that retains eggs into the secondhalf of development is faced not only with the problem oftransporting a clutch that is increasing in mass but alsowith the potential problem of passing enlarged eggsthrough the pelvic girdle during oviposition (Sinervo andLicht 1991) The rarity of oviposition after stage 34 maytherefore be a direct consequence of increasing egg size

Our concern in this article is largely with the proximatephysiological factors that constrain the embryonic stagesat which eggs are oviposited However it is also importantto recognize that many ultimate factors affect the stage atoviposition For example extended egg retention mayhave a high survival cost for some species because of theburden that the clutch imposes on the female (Sinervo andAdolph 1991) Such a cost would favor oviposition as soonas the shell is formed (as long as suitable nest sites areavailable) Clutch frequency should also affect the stage atoviposition Species that produce multiple clutches perseason could minimize the interval between clutches bylaying eggs at early stages of development thus producingmore clutches within a breeding season On the otherhand for species that are limited to one clutch per seasonthe length of time that eggs spend in the oviduct wouldnot reduce the reproductive output of the female Ifembryos continue to develop in the oviduct in suchspecies then delayed oviposition would shorten the lengthof time that the eggs are in the nest and vulnerable topredators or other mortality agents Single-clutchedspecies would thus be more likely than multiple-clutchedspecies to oviposit when embryos are at advanced stages

This prediction is borne out by data on lizards (Figure3) few multiple-clutched lizard species oviposit later thanstage 30 whereas most single-clutched lizard speciesoviposit at stage 30 or later (X2 = 70 P lt 001 2 acute 2 con-tingency analysis two-tailed test) Despite the associationbetween clutch frequency and stage at oviposition webelieve that selection for oviposition to occur at very earlyor very late stages of embryonic development will seldombe effective because of the time necessary to complete theegg shell on the one hand and physiological constraintson embryonic development in utero on the other

Consequences of extended egg retentionfor sceloporine lizardsOur research focuses on spiny lizards of the genus Scelo-porus and tree lizards of the genus Urosaurus Sceloporusand Urosaurus are sister taxa that is each otherrsquos closestrelative Urosaurus contains approximately 10 species allof which are oviparous Sceloporus contains approximate-ly 70 species almost half of which are viviparous (Sites etal 1992) Viviparity has apparently arisen only four timesin this genus once in the ancestor of the formosus grouponce in the ancestor of the grammicus megalepidurus andtorquatus groups and twice within the scalaris group

(Meacutendez-de la Cruz et al 1998) The scalaris group is theonly one to have both oviparous and viviparous speciesand the close relationship of the oviparous and viviparousspecies suggests that viviparity is of recent origin in thisgroup (Benabib et al 1997) The few origins of viviparityand the large number of species groups (approximately15) in which all members are oviparous suggests that thetransition to viviparity may be difficult in Sceloporus

Because oviposition at successively later stages ofembryonic development is the putative transition toviviparity the present-day consequences of extended eggretention on embryonic development provide insightsinto how easy it is for viviparity to evolve We have there-fore studied the consequences of the retention of eggsbeyond the normal stage of oviposition We initiallyfocused on oviparous species known to retain eggs toadvanced stages but a more complete picture requiredthat we also look at the consequences of egg retention bythe more typical speciesmdashthat is those that oviposit whenembryos reach approximately stage 30

The experimental protocol common to all of our stud-ies has been to keep gravid females of each species in thelaboratory under conditions that inhibit oviposition Thisapproach allowed us to assess the capacity of females tosupport embryonic development in utero past the time ofnormal oviposition To inhibit oviposition we tookadvantage of a normal response to drought because suc-cessful nesting in the field is dependent on adequate soilmoisture and rainfall is often unpredictable femalesrespond to drought by retaining their eggs (Andrews andRose 1994) Experimental females were thus kept in cageswith dry soil as a substrate which simulated drought con-ditions and inhibited oviposition (Andrews and Rose1994) Control females were housed under the same con-ditions as experimental females except that the cage sub-strate was moist soil most females oviposited normallyunder these conditions Eggs from control females wereincubated in a standard incubation medium and at thesame temperature as eggs within experimental females

One set of experiments examined egg retention bySceloporus scalaris an oviparous species that is a model forthe transition from oviparity to viviparity (Figure 4) Sscalaris a member of the scalaris species group occurs ingrasslands at many elevations in Arizona New Mexicoand northern Mexico Females from a low-elevation pop-ulation are able to retain eggs for an unprecedented lengthof time although they normally oviposit at stages 31ndash33(Mathies and Andrews 1995 1996 Thomas Mathies andRobin M Andrews unpublished data) Eggs retained aslong as 1 month (to stage 38) and surgically removed fromfemales develop at the same rate and have the same sur-vival to hatching rate as control eggs Eggs retained for afurther 10ndash13 days advance to stage 40 these eggs pro-duced viable hatchlings 1ndash3 days after they were surgicallyremoved from females When the same experiment wasconducted at a slightly higher incubation temperature

232 BioScience March 2000 Vol 50 No 3

Articles

development was somewhat slowerfor retained embryos than for con-trol embryos presumably becausethe higher incubation temperatureincreased the oxygen requirementsof embryos beyond that availablein utero (Andrews 1997)

Nevertheless among squamatesS scalaris has an unusual ability toextend egg retention to within afew days of hatching and to do sowith little or no impairment ofembryonic development The abil-ity to support embryogenesis inutero to advanced stages is alsofound in another oviparous speciesin the scalaris group Sceloporusaeneus (Robin M Andrewsunpublished observation) the sis-ter species of the viviparous Scelo-porus bicanthalis The ability ofoviparous species in the scalarisgroup to extend egg retention toadvanced embryonic stages has likely facilitated the evolu-tion of viviparity in this group

Sceloporus virgatus an oviparous member of the undu-latus species group also normally lays eggs at stages 31ndash33In contrast to S scalaris however when eggs of S virgatusare retained past the normal time of oviposition as long asto stage 37 embryonic development is retarded (Figure 4Andrews and Rose 1994 Andrews 1997) For example thedry masses of embryos retained for 30 days past the nor-mal time of oviposition were less than half those of con-trols and retained embryos showed considerably lowersurvival to hatching than control eggs Female S virgatusthat retain eggs beyond the normal time of ovipositionthus have greatly reduced fitness This assertion is not toimply that extended egg retention has no benefit in thisspeciesmdashfemales that do not retain eggs until the seasonalrains occur would have zero fitness for that reproductiveseason

S virgatus appears to be unusual both within the undu-latus species group and within the genus Sceloporus in itsability to normally support embryonic development paststage 30 (Mathies 1998) No other species that we studiedin the undulatus species group is able to retain eggs appre-ciably beyond stage 30 For example females of Sceloporusoccidentalis another member of the undulatus groupoviposit when embryos reach stages 28ndash30 even if condi-tions are not suitable for nesting This pattern is also char-acteristic of Sceloporus graciosus (a member of the gracio-sus species group) Females of two members of theundulatus species groupmdashSceloporus undulatus consobri-nus from Arizona and Sceloporus undulatus hyacinthinusfrom Virginia and Missourimdashboth typically oviposit whenembryos are at stage 28ndash30 (Sexton and Marion 1974

Mathies 1998) as do female Sceloporus clarki (a member ofthe clarki species group) Moreover embryos of thesespecies do not develop further when eggs are retainedbeyond the normal time of oviposition (Mathies 1998)Initially the similar stages at oviposition for S u consobri-nus and S u hyacinthinus housed on moist and dry sub-strates would appear to be the result of the brief periodapproximately 10 days of facultative egg retention underdrought conditions However during this 10-day periodembryos in control eggs had reached stage 32 and had drymasses four to eight times greater than those of theretained embryos These observations indicate that devel-opment of retained embryos was arrested Females in theundulatus graciosus and clarki species groups thus exhib-it a limited ability to extend egg retention and to supportembryogenesis in utero beyond stage 30

In contrast to members of the undulatus group a mem-ber of the sister genus to Sceloporus Urosaurus ornatus isable to retain eggs for at least a month past the time ofnormal oviposition (Mathies and Andrews 1999) As inmost members of the undulatus group however develop-ment is arrested at approximately the time of normal ovipo-sition In this case development is arrested at stage 305whereas control embryos reach stage 38 after the samelength of incubation (Figure 4) When retained eggs arelaid development is reinitiated and continues to hatchingHatching success of retained eggs is as high as that of con-trol eggs Retention in Urosaurus thus has no obvious effecton embryonic development other than to delay hatchingand then only by the time spent in arrest Retention doesaffect hatchlings however Hatchlings produced after longperiods of retention are smaller and less hydrated thanhatchlings produced after short periods of retention

March 2000 Vol 50 No 3 BioScience 233

Articles

Figure 4 Embryo stage in three species of phrynosomatid lizards as a function of dayselapsed since the normal time of oviposition Eggs of control females (open circles)were oviposited at the normal time Eggs of experimental females (solid circles) wereretained in the oviducts for the number of days indicated Oviposition byexperimental females was inhibited by providing dry substrates for nesting Thedevelopment of the retained embryos was arrested for Urosaurus ornatus (left) wasretarded for Sceloporus virgatus (middle) and proceeded normally for Sceloporusscalaris (right) Figure modified from Andrews and Rose (1994) Mathies and Andrews(1996) and Mathies (1998)

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

development was somewhat slowerfor retained embryos than for con-trol embryos presumably becausethe higher incubation temperatureincreased the oxygen requirementsof embryos beyond that availablein utero (Andrews 1997)

Nevertheless among squamatesS scalaris has an unusual ability toextend egg retention to within afew days of hatching and to do sowith little or no impairment ofembryonic development The abil-ity to support embryogenesis inutero to advanced stages is alsofound in another oviparous speciesin the scalaris group Sceloporusaeneus (Robin M Andrewsunpublished observation) the sis-ter species of the viviparous Scelo-porus bicanthalis The ability ofoviparous species in the scalarisgroup to extend egg retention toadvanced embryonic stages has likely facilitated the evolu-tion of viviparity in this group

Sceloporus virgatus an oviparous member of the undu-latus species group also normally lays eggs at stages 31ndash33In contrast to S scalaris however when eggs of S virgatusare retained past the normal time of oviposition as long asto stage 37 embryonic development is retarded (Figure 4Andrews and Rose 1994 Andrews 1997) For example thedry masses of embryos retained for 30 days past the nor-mal time of oviposition were less than half those of con-trols and retained embryos showed considerably lowersurvival to hatching than control eggs Female S virgatusthat retain eggs beyond the normal time of ovipositionthus have greatly reduced fitness This assertion is not toimply that extended egg retention has no benefit in thisspeciesmdashfemales that do not retain eggs until the seasonalrains occur would have zero fitness for that reproductiveseason

S virgatus appears to be unusual both within the undu-latus species group and within the genus Sceloporus in itsability to normally support embryonic development paststage 30 (Mathies 1998) No other species that we studiedin the undulatus species group is able to retain eggs appre-ciably beyond stage 30 For example females of Sceloporusoccidentalis another member of the undulatus groupoviposit when embryos reach stages 28ndash30 even if condi-tions are not suitable for nesting This pattern is also char-acteristic of Sceloporus graciosus (a member of the gracio-sus species group) Females of two members of theundulatus species groupmdashSceloporus undulatus consobri-nus from Arizona and Sceloporus undulatus hyacinthinusfrom Virginia and Missourimdashboth typically oviposit whenembryos are at stage 28ndash30 (Sexton and Marion 1974

Mathies 1998) as do female Sceloporus clarki (a member ofthe clarki species group) Moreover embryos of thesespecies do not develop further when eggs are retainedbeyond the normal time of oviposition (Mathies 1998)Initially the similar stages at oviposition for S u consobri-nus and S u hyacinthinus housed on moist and dry sub-strates would appear to be the result of the brief periodapproximately 10 days of facultative egg retention underdrought conditions However during this 10-day periodembryos in control eggs had reached stage 32 and had drymasses four to eight times greater than those of theretained embryos These observations indicate that devel-opment of retained embryos was arrested Females in theundulatus graciosus and clarki species groups thus exhib-it a limited ability to extend egg retention and to supportembryogenesis in utero beyond stage 30

In contrast to members of the undulatus group a mem-ber of the sister genus to Sceloporus Urosaurus ornatus isable to retain eggs for at least a month past the time ofnormal oviposition (Mathies and Andrews 1999) As inmost members of the undulatus group however develop-ment is arrested at approximately the time of normal ovipo-sition In this case development is arrested at stage 305whereas control embryos reach stage 38 after the samelength of incubation (Figure 4) When retained eggs arelaid development is reinitiated and continues to hatchingHatching success of retained eggs is as high as that of con-trol eggs Retention in Urosaurus thus has no obvious effecton embryonic development other than to delay hatchingand then only by the time spent in arrest Retention doesaffect hatchlings however Hatchlings produced after longperiods of retention are smaller and less hydrated thanhatchlings produced after short periods of retention

March 2000 Vol 50 No 3 BioScience 233

Articles

Figure 4 Embryo stage in three species of phrynosomatid lizards as a function of dayselapsed since the normal time of oviposition Eggs of control females (open circles)were oviposited at the normal time Eggs of experimental females (solid circles) wereretained in the oviducts for the number of days indicated Oviposition byexperimental females was inhibited by providing dry substrates for nesting Thedevelopment of the retained embryos was arrested for Urosaurus ornatus (left) wasretarded for Sceloporus virgatus (middle) and proceeded normally for Sceloporusscalaris (right) Figure modified from Andrews and Rose (1994) Mathies and Andrews(1996) and Mathies (1998)

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

These observations on sceloporine lizards are the firstreports of developmental arrest of well-differentiated rep-tilian embryos By contrast turtle and chameleon embryosarrest much earlier in development The survival of well-developed embryos during developmental arrest suggeststhat gas exchange is sufficient to support the maintenancemetabolism of these embryos but not their differentiationor growth These observations also suggest that develop-mental arrest may actually be a common feature of thereproductive biology of oviparous squamates It may nothave been reported before simply because it is notdetectable without experimental manipulation of thelength of retention and the provision of an appropriatecontrol That is a species in which development is arrest-ed at stage 30 will always appear to lay eggs at the ldquorightrdquostage

A third notable finding from these studies is thatextended egg retention actually encompasses two phe-nomena both of which bear on the ease with whichviviparity is likely to evolve (Figure 5) One phenomenonis the ability to maintain gravidity that is to facultativelyretain eggs beyond the time when oviposition normally

occurs This ability is related to the secreto-ry activity of the corpora lutea progesteroneproduced by this structure inhibits uterinecontractions and maintains hyperemia andthe depth of the oviductal mucosa (Guilletteand Jones 1985 Shine and Guillette 1988)The second phenomenon is the ability tosupport embryogenesis in utero past thestage at which oviposition normally occursThe evolution of these two phenomena isnot necessarily linked In some speciesfemales are able to maintain gravidity as wellas support embryogenesis (eg S scalaris)In other species females can maintain gra-vidity but embryonic development is arrest-ed (eg U ornatus)

Viviparity seems least likely to evolve forspecies in which females have a limited abil-ity to maintain gravidity and embryonicdevelopment is arrested even after shortperiods of retention (lower left of Figure 5)In contrast viviparity seems most likely toevolve when the ability to maintain gravidi-ty is combined with continued embryogen-esis in utero (upper right of Figure 5) Inspecies that are able to maintain graviditybut in which embryonic development isarrested the evolution of viviparity mightseem to be limited (lower right of Figure 5)However it could well be that once the abil-ity to maintain gravidity is in place the fur-ther shift to embryonic development inutero and ultimately to viviparity might beeasier than if selection had to act on both

traits simultaneously

Physiological constraints on theevolution of viviparityComparisons among the major groups of reptiles suggeststhat just as viviparity is more likely to evolve in somesceloporine lizard species than in others it may also bemore likely to evolve in some groups than in othersViviparity has evolved at least 100 times within reptiles(Shine 1985) Viviparous species are however distributedunevenly at all taxonomic levels Turtles archosaurs (croco-dilians and birds) and sphenodontidans are entirelyoviparous whereas viviparity has evolved numerous timesamong lizards and snakes Within the squamates somefamilies are entirely oviparous and others are entirely vivi-parous Even within families in which both modes ofreproduction are represented independent origins of vivi-parity are more common in some lineages than in others(Shine 1985) These observations suggest the existence ofconstraints at various levels (phylogenetic physiologicaland ecological) on the evolution of viviparity in reptiles

Our research has emphasized physiological constraints

234 BioScience March 2000 Vol 50 No 3

Articles

Figure 5 The maximum stage that embryos of Sceloporus and Urosauruslizards reach while retained in utero as a function of the time that females areable to retain eggs The time females are able to retain eggs is expressed as thepercentage of the normal length of the incubation period that is fromoviposition to hatching The ability to support embryogenesis in utero canevolve along any trajectory within the shaded area proceeding from thebottom left to the upper right The upper edge of the shaded area representsthe necessary association between the length of egg retention and the stage ofdevelopment for embryos that continue to develop while retained in uteroAbbreviations are as follows Sc Sceloporus undulatus consobrinus SclSceloporus clarki Sg Sceloporus graciosus Sh Sceloporus undulatushyacinthinus So Sceloporus occidentalis Sv Sceloporus virgatus SsSceloporus scalaris Uo Urosaurus ornatus

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

on embryonic development when embryos are retainedwithin the oviducts Determining the consequences ofextended egg retention by oviparous species on embryon-ic development provides a way to evaluate the relative easewith which viviparity could evolve in different lineages(Figure 5) For example viviparity has and could evolvereadily in the scalaris species group Extended egg reten-tion has no obvious limits or disadvantages for embryon-ic development in this group two oviparous members ofthis lineage can extend egg retention to stages 38ndash40 with-out compromising the rate of embryonic development orhatching success In contrast whereas females of S virga-tus in the undulatus species group can extend egg reten-tion to stage 37 (albeit greatly retarding embryonic devel-opment) females of other species in this group are unableto support embryogenesis past stage 30 regardless of envi-ronmental circumstance Moreover because most speciesin the undulatus group exhibit a limited ability to main-tain gravidity continued development in utero even ifphysiologically possible would be precluded These obser-vations suggest that the evolution of viviparity in theundulatus group although possible is much less likelythan in the scalaris group

The arrest or retardation of embryonic development inutero implicates some form of physiological limitationBecause carbon dioxide diffuses more rapidly than oxygenin liquid hypercapnia is less likely to limit embryonicdevelopment than hypoxia Insufficient availability ofwater is also unlikely to be a general constraint on devel-opment in utero Turtle and crocodilian eggs are providedwith all the water the embryos will need for their entiredevelopment yet development in the oviducts is very lim-ited In contrast squamate eggs are provided with mini-mal amounts of water substantial water uptake from thenest substrate is usually necessary for normal develop-ment Nevertheless embryos in the retained eggs of Sscalaris can develop normally almost to hatching with lit-tle additional water uptake (Mathies and Andrews 1996)

We therefore infer that hypoxia is the most likely limit-ing factor for embryonic development in utero turtle andcrocodilian embryos may simply experience hypoxia atearlier developmental stages than lizard embryos Thesupport for this assumption is however largely indirectFor example one line of evidence is that viviparity in rep-tiles is always associated with the thinning or loss of theshell and the consequent proximity of the oviduct and theembryonic membranes a feature that presumably en-hances gas exchange Another line of evidence is thatembryos of turtles and crocodilians that develop in uteropast the stage of normal oviposition die before hatching orare deformed (Ewert 1985 Ferguson 1985)

Three morphological features determine the availabilityof oxygen to embryos these features either alone or incombination could account for differences in the capacityto support embryogenesis in utero between turtles andsquamates and among species of phrynosomatid lizards

One feature is the thickness and structure of the shellEggshells of reptiles differ enormously in the mineraliza-tion of the outer shell layer and in the density organiza-tion and size of the proteinaceous fibrils that constitutethe inner shell layer (Packard and DeMarco 1991) Bothlayers act as barriers to gas diffusion although the fibrillayer may be the more effective (Feder et al 1982) Thesecond feature is the extent of vascularization of the extra-embryonic membranes that line the inner surface of theeggshell These membranes are the yolk sac which is theonly vascularized extra-embryonic membrane early indevelopment and the chorioallantois which becomesincreasingly important for gas exchange as it grows overthe inner surface of the shell during the latter part ofdevelopment In two species of Sceloporus for examplethe chorioallantois covers 20ndash30 of the inner surface ofthe shell by stage 30 and the entire inner surface by stages36ndash38 Coverage by the chorioallantois is completed earli-er in S scalaris than in S virgatus which is consistent withtheir respective abilities to support embryonic develop-ment in utero (Andrews 1997) The third feature is thevascularity of the oviduct Vascularity of the oviduct variesduring the reproductive cycle (Masson and Guillette 1987)and between oviparous and viviparous sister species(Guillette and Jones 1985) The amount of oxygen avail-able for embryonic metabolism is presumably affected bythe proximity of the vascular membranes of the embryo tothe oviduct by the efficiency of oxygen transfer betweenmaternal and fetal capillaries and by the amount of capil-lary development

To assess the importance of gas exchange as a limitingfactor for embryonic development in the oviduct we eval-uated the hypothesis that differences in shell structure areassociated with differences in the capacity of embryos tocontinue development in the oviduct This hypothesis isattractive because viviparity is thought to require a majorreduction of the shell This hypothesis was however notsupported at least for the lizards that we studied (Figure 6Mathies 1998) For example two features of the shell thatshould enhance gas exchangemdashthickness and conduc-tance to water vapormdashare related not to the extent ofembryonic development that can occur in utero but ratherto the mass of the egg It thus appears unlikely that somefeature of the eggshell accounts for differences in theamount of embryonic development that can take place inthe oviduct among species of Sceloporus Hypoxia is stilllikely to be the limiting factor however The eggshell is butone of three morphological features that affect oxygenavailability to embryos the contributions of the two otherfeatures the extra-embryonic membranes and the ovi-duct to the diffusion of gases during development haveyet to be determined for any reptile

Our conclusion that viviparity is more apt to evolve insome groups than in others because of phylogenetic vari-ation in a physiological constraint is based on a small sub-set of reptilian species and we do not know whether the

March 2000 Vol 50 No 3 BioScience 235

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

same constraint operates in other groups The stage atoviposition is known for fewer than 100 of the approxi-mately 5500 species of oviparous squamates and in somefamilies the stage at oviposition is not known for even onerepresentative Moreover the consequences of experimen-tally extended egg retention have been determined foronly a few species of phrynosomatid lizards Neverthelesswhat little information is available reveals a wide diversityof embryonic natural histories Additional research onhow long embryos are retained in the oviduct and howmuch development takes place during this time will sure-ly provide important insights into the evolution of vivi-

parity the information in this article is the tip of the ice-berg with respect to the potential wealth of informationthat awaits discovery

Further understanding of constraints on the evolutionof viviparity would be greatly enhanced by systematicstudies on the natural history of embryos One specificresearch target is groups for which little to no informationis available (eg Amphisbaenia Anguidae CordylidaePygopodidae Varanidae and snakes in general) A secondtarget is groups that show substantial variation in embry-onic natural history (Chamaeleontidae TropiduridaeScincidae Colubridae) Comparative research on closely

236 BioScience March 2000 Vol 50 No 3

Articles

Figure 6 Lack of correlation between eggshell structure and the maximal embryo stage attained in utero in the sister speciesSceloporus undulatus consobrinus and Sceloporus virgatus (a) Edge view of an eggshell of S u consobrinus (b) Edge view ofan eggshell of S virgatus Eggshells examined at the time of oviposition consist of an outer calcareous layer (OS) and an innerlayer of densely packed fibers the shell membrane (SM) Arrows indicate a thin cuticle deposited on the outer surface of theshell Scale bars = 30 m m The eggshells of the two species are indistinguishable in morphology and permeability to water vapor(c and d) The relative sizes and degrees of differentiation that embryos of S u consobrinus and S virgatus reach in uteroBecause embryos of these species were not available they are represented by embryos of Sceloporus scalaris at stage 31 (c) andstage 37 (d) respectively At these stages embryos of all Sceloporus are similar in size and appearance The embryos in (c) and(d) are approximately 24 and 44 mm in maximum dimension respectively Despite the similarity of their eggshells themaximal size and degree of differentiation reached in utero are considerably greater in S virgatus than in S u consobrinus

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

related taxa with diverse embryonic natural histories with-in these families would be especially informative A thirdtarget is those squamates that have atypical embryonicnatural histories For example unlike most squamatesgekkonine and sphaerodactyline geckos lay rigid-shelledeggs and often place them in dry exposed environmentsThese taxa thus have both atypical eggs and atypicaloviposition sites Are eggs of these geckos relatively imper-vious to gas exchange Would their embryos thereforeexperience hypoxia at earlier stages than species of diplo-dactyline and eublepharine geckos which produce flexi-ble-shelled eggs Are their eggs therefore laid when theembryos are at early developmental stages Are exposednests necessary to ensure adequate gas exchange forembryonic development All of these questions would beeasy to address and their answers would provide generalinsights into constraints on egg retention

We have argued that physiological constraints are themost important proximate factors affecting the potentialfor particular taxa to shift from oviparity to viviparityClearly however physiological factors are not the onlyconstraints that have affected the evolution of viviparityFor example the very early stage at oviposition thatappears to be typical of varanid lizards has at least severalpossible explanations Perhaps as active predators vara-nids lay their eggs as soon as possible to avoid detrimentalburdening of the female Alternatively their particularchoice of nest sites may favor sturdy eggs that are resistantto desiccation predators or pathogens but that as a con-sequence have poor gas exchange capabilities Whateverthe ultimate explanation or explanations for the oviparityof varanids in particular the present-day reality is thatobligate oviposition at early embryonic stages by somesquamates turtles crocodilians and sphenodontidanspossibly associated with developmental arrest in uteroprecludes viviparity in these taxa now and would havedone so in the past

AcknowledgmentsWe thank Benoit Heulin Claude Guillaume Fiona Quallsand Rick Shine for their unpublished observations onclutch frequency Support was provided by NSF grant noBSR-9022425 to R M A and by Sigma Xi and TheodoreRoosevelt grants to T M

References citedAndrews RM 1979 Reproductive effort of female Anolis limifrons (Sauria

Iguanidae) Copeia 1979 620ndash626______ 1985 Oviposition frequency of Anolis carolinensis Copeia 1985

259ndash262______ 1997 Evolution of viviparity Variation between two sceloporine

lizards in the ability to extend egg retention Journal of Zoology (Lon-don) 243 579ndash595

Andrews RM Rose BR 1994 Evolution of viviparity Constraints on eggretention Physiological Zoology 67 1006ndash1024

Benabib M Kjer KM Sites JW Jr 1997 Mitochondrial DNA sequence-based phylogeny and the evolution of viviparity in the Sceloporusscalaris group (Reptilia Squamata) Evolution 51 1262ndash1275

Bennett AF 1987 The accomplishments of ecological physiology Pages1ndash8 in Feder ME Bennett AF Burggren WW Huey RB eds New Direc-tions in Ecological Physiology Cambridge (UK) Cambridge Universi-ty Press

Billy AJ 1988 Observations on the embryology of the unisexual lizardCnemidophorus uniparens (Teiidae) Journal of Zoology (London) 21555ndash81

Birchard GG Walsh T Rosscoe R Reiber CL 1995 Oxygen uptake byKomodo dragon (Varanus komodoensis) eggs The energetics of pro-longed development in a reptile Physiological Zoology 68 622ndash633

Black GP Birchard GF Schuett GW Black VD 1984 Influence of incuba-tion water content on oxygen uptake in embryos of the Burmesepython (Python molurus bioittatus) Pages 137ndash145 in Seymour RS edRespiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Blackburn DG 1992 Convergent evolution of viviparity matrotrophy andspecializations for fetal nutrition in reptiles and other vertebratesAmerican Zoologist 32 313ndash321

______1995 Saltationist and punctuated equilibrium models for the evo-lution of viviparity and placentation Journal of Theoretical Biology174 199ndash216

Bons J Bons N 1960 Notes sur la reproduction et le deacuteveloppement deChamaeleo chamaeleon (L) Bulletin Socieacuteteacute des Sciences Naturelles etPhysiques du Maroc 40 323ndash335

Buhlmannx KA Lynch TK Gibbons JW Greene JL 1995 Prolonged eggretention in the turtle Deirochelys reticularia in South Carolina Her-petologica 51 457ndash462

Congdon JD Gibbons JW Greene JL 1983 Parental investment in thechicken turtle (Deirochelys reticularia) Ecology 64 419ndash425

Cree A Cockrem JF Guillette LJ Jr 1992 Reproductive cycles of male andfemale tuatara (Sphenodon punctatus) on Stephens Island NewZealand Journal of Zoology (London) 226 199ndash217

DeMarco V 1993 Estimating egg retention times in sceloporine lizardsJournal of Herpetology 27 453ndash458

Dufaure JP Hubert J 1961 Table de deacuteveloppement du leacutezard vivipareLacerta (Zootoca) vivipara Jacquin Archives Anatomie MicroscopieMorphologie Expeacuterimental 50 309ndash328

Ewert MA 1985 Embryology of turtles Pages 75ndash267 in Gans C Billett FMaderson PFA eds Biology of the Reptilia Vol 14 Development ANew York John Wiley amp Sons

Feder ME Satel SL Gibbs AG 1982 Resistance of the shell membrane andthe mineral layer to diffusion of oxygen and water in flexible-shelledeggs of the snapping turtle (Chelydra serpentina) Respiratory Physiol-ogy 49 279ndash291

Ferguson MWJ 1985 Reproductive biology and embryology of the croco-dilians Pages 329ndash491 in Gans C Billett F Maderson PFA eds Biolo-gy of the Reptilia Vol 14 Development A New York John Wiley ampSons

Guillette LJ Jr 1982 The evolution of viviparity and placentation in thehigh elevation Mexican lizard Sceloporus aeneus Herpetologica 3894ndash103

______ 1992 Morphology of the reproductive tract in a lizard exhibitingincipient viviparity (Sphenomorphus fragilis) and its implications forthe evolution of the reptilian placenta Journal of Morphology 212163ndash173

______ 1993 The evolution of viviparity in lizards BioScience 43742ndash751

Guillette LJ Jr Jones RE 1985 Ovarian oviductal and placental morphol-ogy of the reproductively bimodal lizard Sceloporus aeneus Journal ofMorphology 184 85ndash98

Heulin B 1990 Etude comparative de la membrane coquilleacutere chez lessouches ovipare et vivipare du leacutezard Lacerta vivipara Canadian Jour-nal of Zoology 68 1015ndash1019

Kennett R Georges A Palmer-Allen M 1993 Early developmental arrestduring immersion of eggs of a tropical freshwater turtle Chelodinarugosa (Testudinata Chelidae) from Northern Australia AustralianJournal of Zoology 41 37ndash45

March 2000 Vol 50 No 3 BioScience 237

Articles

Masson GR Guillette LJ Jr 1987 Changes in oviductal vascularity duringthe reproductive cycle of three oviparous lizards (Eumeces obsoletusSceloporus undulatus and Crotaphytus collaris) Journal of Reproduc-tion and Fertility 80 361ndash372

Mathies T 1998 Constraints on the evolution of viviparity in the lizardgenus Sceloporus PhD dissertation Virginia Polytechnic Institute andState University Blacksburg VA

Mathies T Andrews RM 1995 Thermal and reproductive ecology of highand low elevation populations of the lizard Sceloporus scalaris Impli-cations for the evolution of viviparity Oecologia 104 101ndash111

______ 1996 Extended egg retention and its influence on embryonicdevelopment and egg water balance Implications for the evolution ofviviparity Physiological Zoology 69 1021ndash1035

______1999 Determinants of embryonic stage at oviposition in the lizardUrosaurus ornatus Physiological and Biochemical Zoology 72 645ndash 655

Meacutendez-de la Cruz FR Villagraacuten-Santa Cruz M Andrews RM 1998 Evo-lution of viviparity in the lizard genus Sceloporus Herpetologica 54521ndash532

Miller JD 1985 Embryology of marine turtles Pages 269ndash328 in Gans CBillett F Maderson PFA eds Biology of the Reptilia Vol 14 Develop-ment A New York John Wiley amp Sons

Moffat LA 1985 Embryonic development and aspects of reproductivebiology in the tuatara Sphenodon punctatus Pages 493ndash521 in Gans CBillett F Maderson PFA eds Biology of the Reptilia Vol 14 Develop-ment A New York John Wiley amp Sons

Packard GC Tracy CR Roth JJ 1977 The physiological ecology of reptil-ian eggs and embryos and the evolution of viviparity within the classReptilia Biological Review 52 71ndash105

Packard GC et al 1989 How are reproductive systems integrated and howhas viviparity evolved Pages 281ndash293 in Wake DB Roth G eds Com-plex Organismal Functions Integration and Evolution in VertebratesNew York John Wiley amp Sons

Packard MJ DeMarco VG 1991 Eggshell structure and formation in eggsof oviparous reptiles Pages 53ndash69 in Deeming DC Ferguson MWJeds Egg Incubation Its Effects on Embryonic Development in Birdsand Reptiles Cambridge (UK) Cambridge University Press

Packard MJ Packard GC 1984 Comparative aspects of calcium metabo-lism in embryonic reptiles and birds Pages 155ndash179 in Seymour RSed Respiration and Metabolism of Embryonic Vertebrates Dordrecht(The Netherlands) Dr W Junk Publishers

Palmer BD DeMarco VG Guillette LJ Jr 1993 Oviductal morphology andeggshell formation in the lizard Sceloporus woodi Journal of Morphol-ogy 217 205ndash217

Phillips JA Millar RP 1998 Reproductive biology of the white-throated

savanna monitor Varanus albigularis Journal of Herpetology 32366ndash377

Qualls CP 1996 Influence of the evolution of viviparity on eggshell mor-phology in the lizard Lerista bougainvillii Journal of Morphology 228119ndash125

Sexton OJ Claypool L 1978 Nest sites of a northern population of anoviparous snake Opheodrys vernalis (Serpentes Colubridae) Journalof Natural History 12 365ndash370

Sexton OJ Marion KR 1974 Duration of incubation of Sceloporus undu-latus eggs at constant temperature Physiological Zoology 47 91ndash98

Shine R 1983 Reptilian reproductive modes The oviparityndashviviparitycontinuum Herpetologica 39 1ndash8

______1985 The evolution of viviparity in reptiles An ecological analysisPages 605ndash694 in Gans C Billett F eds Biology of the Reptilia Vol 15Development B New York John Wiley amp Sons

Shine R Guillette LJ Jr 1988 The evolution of viviparity in reptiles Aphysiological model and its ecological consequences Journal of Theo-retical Biology 132 43ndash50

Sinervo B Adolph SC 1991 Decreased sprint speed as a cost of reproduc-tion in the lizard Sceloporus occidentalis Variation among populationsJournal of Experimental Biology 155 323ndash336

Sinervo B Licht P 1991 Proximate constraints on the evolution of egg sizenumber and total clutch mass in lizards Science 252 1300ndash1302

Sites JW Jr Archie JW Cole CJ Flores-Villela O 1992 A review of phylo-genetic hypotheses for lizards of the genus Sceloporus (Phrynosomati-dae) Implications for ecological and evolutionary studies Bulletin ofthe American Museum of Natural History 213 1ndash110

Smith SA Shine R 1997 Intraspecific variation in reproductive modewithin the scincid lizard Saiphos equalis Australian Journal of Zoology45 435ndash445

Thompson MB 1985 Functional significance of the opaque white patch ineggs of Emydura macquarii Pages 387ndash395 in Grigg G Shine REhrmann H eds Biology of Australian Frogs and Reptiles ChippingNorton (Australia) Surrey Beatty and Sons

Weekes HC 1935 A review of placentation among reptiles with particularregard to the function and evolution of the placentaProceedings of theZoological Society of London 2 625ndash645

Wourms JP Lombardi J 1992 Reflections on the evolution of piscineviviparity American Zoologist 32 276ndash293

Xavier F Gavaud J 1986 Oviparityndashviviparity continuum in reptiles Phys-iological characteristics and relation with environment Pages 79ndash93 inAssenmacher I Bossin J eds Endocrine Regulations as Adaptive Mech-anisms to the Environment Paris Centre National de la Recherche Sci-entifique

238 BioScience March 2000 Vol 50 No 3

Articles