Incubation Regimes of Cold-Climate Reptiles: the Thermal Consequences of Nest-Site Choice, Viviparity and Maternal Basking

Blackwell Science, LtdOxford, UKBIJBiological Journal of the Linnean Society0024-4066The Linnean Society of London, 2004? 2004 832 145155 Original Article

INCUBATION REGIMES IN COLD-CLIMATE REPTILES R. SHINE

Biological Journal of the Linnean Society, 2004, 83, 145–155. With 4 figures

Incubation regimes of cold-climate reptiles: the thermal consequences of nest-site choice, viviparity and maternal basking

RICHARD SHINE*

Biological Sciences A08, University of Sydney, NSW 2006 Australia

Received 23 October 2003; accepted for publication 19 January 2004

Cold-climate reptiles show three kinds of adaptation to provide warmer incubation regimes for their developing embryos: maternal selection of hot nest sites; prolonged uterine retention of eggs; and increased maternal basking during pregnancy. These traits may evolve sequentially as an oviparous lineage invades colder climates. To compare the thermal consequences of these adaptations, I measured microhabitat temperatures of potential nest sites and actual nests of oviparous scincid lizards (Bassiana duperreyi), and body temperatures of pregnant and non-pregnant viviparous scincid lizards (Eulamprus heatwolei). These comparisons were made at a site where both species occur, but close to the upper elevational limit for oviparous reptiles in south-eastern Australia. Viviparity and maternal basking effort had less effect on mean incubation temperature than did maternal nest-site selection. Eggs retained in utero experienced bimodal rather than unimodal diel thermal distributions, but similar mean incubation temperatures. Often the published literature emphasizes the ability of heliothermic (basking) reptiles to maintain high body temperatures despite unfavourable ambient weather conditions; this putative ability is central to many hypo- theses on selective forces for the evolution of viviparity. In cold climates, however, opportunities for maternal thermoregulation to elevate mean body temperatures (and thus, incubation temperatures) above ambient levels may be severely limited. Hence, at least over the broad elevational range in which oviparous and viviparous species live in sympatry, maternal selection of ‘hot’ nests may be as effective as is viviparity in providing favourable incubation regimes. © 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 83, 145–155.

ADDITIONAL KEYWORDS: embryo – lizard – reproductive mode – thermoregulation.

INTRODUCTION ticular types of organism. Such limits will involve more than simple physical parameters; biology must Changes in global climate have focused attention on also be considered, because organisms at the upper thermally imposed limits to the geographical distribu- elevational limit of their distribution will be under tions of fauna and ﬂora, because the spatial locations strong selective pressures to expand their ranges of such limits will shift dramatically with global either by physiological adaptations to cooler condi- warming. Such changes are likely to be most impor- tions or by behavioural selection of warmer-than- tant for taxa in montane areas, where even a small average microhabitats (Shine, 1999; Shine, Barrott & rise in mean temperature may enable lowland ani- Elphick, 2002a; Langkilde, O’Connor & Shine, 2003). mals and plants to penetrate much further up the Quantifying the effect of the resulting adaptations is mountains (e.g. Hughes, Cawsey & Westoby, 1996a, b; an important step towards clarifying the processes Gibbons et al., 2000). The ﬁrst step in predicting the that determine distributional limits for montane magnitude of such effects is to understand exactly how species. thermal factors limit elevational distribution for par- For many kinds of animal, life-history stages vary in their vulnerability to thermal conditions. Often, the most vulnerable stage is the embryo. Because it is *E-mail: [email protected] immobile, at least in oviparous (egg-laying) species

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 83, 145–155 145 146 R. SHINE where eggs are deposited in a nest and remain there or more stable temperatures than would be experi- throughout development, its thermoregulatory ability enced under the thermal regimes typical of non- is limited. In contrast, free-living juveniles and pregnant conspecifics (Beuchat, 1986; Rosen, 1991; adults can select thermally optimal microhabitats, Tu & Hutchinson, 1994; Andrews, DelaCruz & and hence can exploit areas that would be lethal SantaCruz, 1997; Rock, Cree & Andrews, 2002; for a sedentary egg. Squamate reptiles (lizards and Rock & Cree, 2003). snakes) provide excellent model systems in which to examine elevational limits to distribution, and the These phenomena suggest that the thermal envi- role of maternal adaptations in extending those lim- ronments experienced by developing reptile embryos its. Incubation temperatures are critical variables in in cold climates may be substantially warmer com- the life history of reptiles: they strongly affect devel- pared with those that would be experienced if females opmental rates (and thus, total incubation periods) of simply laid eggs at random among potential nest sites. eggs, and also may affect many phenotypic traits Importantly, the maternal adaptations outlined above (size, shape, sex, locomotor speeds) of the resultant are likely to accumulate sequentially within any given hatchlings (Burger, 1989; Warner & Andrews, 2002). lineage of cold-climate reptiles. That is, the first ovi- In turn, the date of hatching and the phenotype of parous females encountering incubation conditions the offspring may affect its probability of survival cold enough to decrease hatchling viability would have (Olsson & Shine, 1997; Warner & Andrews, 2002). been under selection to choose warmer-than-usual Reflecting this central role for incubation tempera- nest sites. In even colder areas, however, thermally ture, thermal tolerances of embryos appear to limit adequate sites become so rare that maternal retention the cold-climate edge of species distributions, at least of developing offspring offers advantages over early in oviparous reptiles. For example, geographical oviposition; and eventually, in colder and colder areas, range limits of oviparous snakes and lizards conform this prolongation extends all the way through to par- to predictions from embryonic thermal tolerances and turition (i.e. viviparity). During or after this transi- mean soil temperatures (Shine, 1987; Qualls et al., tion, females that modify their thermoregulatory 1996; Qualls, 1997). behaviour to further ameliorate embryonic tempera- Cold-climate reptiles display three main types of tures produce earlier and/or fitter offspring. Thus, adaptation that allow their embryos to develop at rel- although intuition suggests that embryos inside a atively high temperatures despite the low ambient viviparous female reptile will be kept substantially thermal regimes: warmer compared with mean soil temperatures in any cold-climate environment, this disparity is actually 1. Selection of warm nest sites. Some cool-climate the summed result of at least three sets of adapta- oviparous squamates lay their eggs in sites that are tions: those for maternal nest-site selection, those for no warmer than are random sites in the surround- prolonged uterine retention, and those for behavioural ing area [e.g. Nannoscincus maccoyi (Shine, 1999)]. thermoregulation of pregnant females. In order to However, females of many cool-climate oviparous evaluate the degree to which reptilian viviparity species actively select warm nests, usually by changes the thermal experience of eggs, the question choosing areas exposed to high levels of solar radi- must be placed within this broader context. That is, we ation and then placing the eggs close to the soil need to consider the impact of viviparity relative to surface (Sexton & Claypool, 1978; Andrews, 2000; maternal behavioural adaptations (nest-site selection Shine et al., 2002a). and increased basking effort) that also affect egg 2. Viviparity. Live-bearing has evolved independently temperatures. in many lineages of oviparous reptiles invading We can estimate the magnitude in thermal shifts cold climates, presumably via progressive increases due to each of these processes by measuring four sets in the duration of uterine retention of developing of thermal regimes: (a) inside a random subset of embryos (Blackburn, 1982, 1985; Shine, 1985). The potential nest sites, (b) inside actual nests, (c) inside primary selective force for this transition appears non-pregnant viviparous lizards, and (d) inside preg- to have been the thermal environment of the nant viviparous lizards. The disparity between (a) and embryos; maternal thermoregulation allows the (b) reflects the effect of maternal nest-site selection, eggs to be kept at higher temperatures than are that between (b) and (c) reflects the effect due to the available in any nest (Shine, 1985, 2002a; but see evolution of viviparity, and that between (c) and (d) is Andrews, 2000). due to pregnancy-induced modifications of thermo- 3. Increased basking effort. In many viviparous spe- regulatory behaviour. What has been the relative cies, pregnancy modifies female thermoregulatory importance of each of these processes to incubation behaviour: pregnant animals bask for longer peri- conditions of cold-climate reptiles? To answer this ods, and thus expose their embryos to higher and/ question, I measured these four sets of thermal

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 83, 145–155 INCUBATION REGIMES IN COLD-CLIMATE REPTILES 147 regimes, at a cold-climate site near the upper eleva- entrecasteauxii, P. pagenstecheri, P. spenceri, Tiliqua tional limit for oviparous reptiles in south-eastern scincoides; the agamid lizard Tympanocryptis diemen- Australia. The sympatry of oviparous and viviparous sis; the elapid snakes Austrelaps ramsayi, Drysdalia species at this site allowed me to make direct compar- coronoides, Pseudechis porphyriacus, Pseudonaja tex- isons between these two reproductive modes. tilis) are heliothermic and exhibit relatively similar mean temperatures during activity (typically around MATERIAL AND METHODS 30 ∞C, Shine, 1983; Greer, 1989, 1997). Also, most oviparous squamate species in this area select exposed, STUDY SPECIES AND AREA relatively hot nests (e.g. Lampropholis delicata, The Brindabella Range, 40 km west of Canberra in the L. guichenoti, Tympanocryptis diemensis, Shine & Australian Capital Territory, consists of eucalypt Harlow, 1996; Shine, Elphick & Harlow, 1997; pers. woodland with occasional open clearings generated by observ.) and communal nests often contain eggs from poor soils, fires, windstorms, or anthropogenic activi- two or three scincid species (Shine, 1983; pers. ties (roads, electric power lines, ski runs). The area observ.). The sole exception in these respects is the fos- experiences cold winters (July mean temperature, sorial skink N. maccoyi: this species and its nests are 10.4 ∞C) and warm summers (January mean temper- restricted to cool damp microhabitats (Robertson, ature, 25.9 ∞C; Australian Bureau of Meteorology). 1981; Shine, 1999). Hence, nest and body tempera- Scincid lizards are abundant in the region, and have tures measured for Bassiana and Eulamprus in this attracted substantial ecological study (Pengilley, 1972; study are likely to be broadly representative of those Shine & Elphick, 2001; Shine et al., 2002a; Shine, exhibited by all of the Brindabella squamate species 2002a, b, 2004; Shine, Elphick & Donnellan, 2002b; except for N. maccoyi. Shine, Elphick & Barrott, 2003). Oviparous lizards extend up to elevations of around 1600 m, whereas METHODS viviparous species extend much further (to >1800 m, To quantify thermal regimes I used thermochron ibut- Pengilley, 1972). Thus oviparous and viviparous taxa tons (Dallas Semiconductor, Dallas, Texas, USA; diam- are broadly sympatric over lower elevations, but only eter 15 mm, height 6 mm, mass 3.3 g) set to record viviparous species occur above 1600 m. This study temperatures every 10 min for 2 weeks. I measured focused on lizards at the base of Mount Ginini [1615 m four sets of thermal regimes: above sea level (a.s.l.); 148∞46¢E, 35∞32¢S]; the oviparous Bassiana duperreyi (= ‘Leiolopisma trilineata’ in 1. Natural nests. In the Brindabella Range, earlier literature) nests at the base of a ski run in this B. duperreyi oviposit selectively in open areas area but we have never found any nests further up the exposed to high levels of solar radiation (Shine mountain, despite the availability of similarly open et al., 2002a). Three people searched for lizard eggs habitat (Shine et al., 2003). Translocation of Bassiana at Mount Ginini in early December 2002, about eggs to artificial nests at these higher sites yielded 1 week after the highly synchronized oviposition of high rates of embryonic mortality (Shine, 2002a); Bassiana in this area (Shine & Elphick, 2001). We thus, both experimental and descriptive data suggest placed thermochrons in each of 15 nests, all of that the site is close to the upper elevational limit for which were under rocks in open clearings. oviparous reproduction. 2. Potential nest sites. Searches, at the same time as A viviparous skink, Eulamprus heatwolei, is also those described above, under rocks and logs in the abundant on Mount Ginini. It is similar to surrounding forests did not reveal any lizard nests. B. duperreyi in most aspects except for reproductive However, previous work has shown that one of the mode. For example, both species have a maximum oviparous scincids in the Brindabella Range, adult snout–vent length of around 80–100 mm, are N. maccoyi, occurs in this area, and oviposits in just diurnally active insectivores, mature at around 2– such sites at slightly lower elevations (Shine, 3 years of age, and adult females produce a single 1999). We placed thermochrons under rocks clutch or litter of around two to six offspring in sum- (n = 10) and logs (n = 2) in representative forested mer each year (Pengilley, 1972; Greer, 1989). Both spe- areas <30 m from actual nests. Because eucalypt cies are heliotherms, with active individuals selecting forest covers >95% of the habitat, this sample of body temperatures of around 30 ∞C (Shine, 1983; ‘nests’ provided an approximation of nests laid ran- Greer, 1989). I selected these two species for study domly with respect to vegetation cover. because of these extensive similarities. However, most 3. Body temperatures of non-pregnant lizards. I col- squamate taxa living in the Brindabella Range lected five male and six non-pregnant female (e.g. the scincid lizards B. duperreyi, Egernia E. heatwolei from surrounding areas; sex was montana, Eg. whitii, E. heatwolei and E. tympanum, assessed by hemipene eversion. I glued thermo- Lampropholis delicata, L. guichenoti, Pseudemoia chrons (with external casings removed to reduce

mass, Robert & Thompson, 2003) on their mid- A dorsal surfaces. Each thermochron weighed 1.5 g, 30 <10% of the mass of the lizard carrying it. Lizards showed no overt reaction to the thermochrons, and 25 continued to move about and bask normally. Cali- bration studies show that the external tempera- 20 tures measured by these units are very similar to internal body temperatures as measured by cloacal 15 probes (Robert & Thompson, 2003). We distributed

these lizards among four open-topped circular are- 10 nas (1.3 m diameter, with 50-cm high walls). Each arena contained natural substrates (rocks, logs, 5 soil, bark), and mesh stretched across the top pre- C)

vented predation by birds. Food and water were ° 0 provided ad libitum. Each arena was exposed to full 0 24681012141618202224 sun for most of the day, but with shade from over- B hanging trees available also. 32

4. Body temperatures of pregnant lizards. Eight preg- 30 nant female E. heatwolei were collected at the same Temperature ( time as were the non-reproductive animals (above), 28 and placed into the same enclosures. Sex was 26 assessed by hemipene eversion and pregnancy by 24 abdominal palpation. Pregnancy was later con- firmed by maintaining these animals in captivity 22 until parturition. At the end of the thermal study 20 the thermochrons were removed, and the lizards 18 were maintained in captivity until parturition; they were later released unharmed at their original 16 capture sites. The E. heatwolei were released into 14 the enclosures on 7 December 2002; the initial plan 0246810121416 18 20 22 24 was to monitor temperatures for 2 weeks, but an Time of day (h) intense wildfire swept through the site on 13 Figure 1. Diel shifts in mean hourly temperatures for four December. The lizards were thus removed on 12 types of potential incubation environment for lizard December, 5 days after the study began. embryos at a high-elevation site at Mount Ginini in the To allow E. heatwolei to acclimate to their enclo- Brindabella Range. The graph shows mean values plus sures, I discarded thermal data from the first day of associated standard errors for potential nest sites in the the experiment. This left 4.5 days of data with records forest (), for actual nests of the oviparous scincid lizard of temperatures at 10-min intervals from 19 lizards, Bassiana duperreyi in open clearings ( ), and for body 15 actual nests, and 12 potential nest sites (total of temperatures of male ( ), non-reproductive female ( ) and 32 732 readings). Statistical analyses of these data pregnant female ( ) viviparous scincid lizards Eulamprus were performed using the program Statview 5 (SAS heatwolei. Each hourly mean value is based on overall mean values for each of the replicate sites in each category, Institute, 1998) on a Macintosh G5 computer. Assump- averaged over the 4 days of monitoring; for example, each tions of relevant statistical tests (normality, homoge- nest contributed only a single value for each hour of the neity of variances) were checked prior to analysis. The day. (A) shows mean values calculated from raw data, and text reports mean values one standard error. ± (B) shows calculations after all values <16.5 ∞C (the lower thermal limit for embryogenesis in B. duperreyi) were replaced with a value of 16.5 ∞C. Thus, (B) shows the effec- RESULTS tive mean incubation temperature, allowing for the cessa- The four thermal regimes differed considerably both tion of development below this threshold level. in mean temperature and in pattern of diel variation in temperature (Fig. 1), and thus in their predicted each other: potential nest sites in the forest consequences for rates of embryonic development: (14.91 ± 0.16 ∞C), male E. heatwolei (14.64 ± 0.61 ∞C), (a) Mean temperatures were higher for actual non-pregnant female E. heatwolei (14.38 ± 0.62 ∞C), B. duperreyi nests (17.27 ± 0.23 ∞C) than they were and pregnant female E. heatwolei (14.45 ± 0.68 ∞C) for other regimes, all of which were very similar to (one-factor ANOVA using mean values per site,

F4,45 = 13.99, P < 0.0001; Fisher’s PLSD posthoc tests A showed that actual nests were hotter than all others 100 at P < 0.0001). (b) Diel patterns of body temperature showed even stronger divergence. Nest sites (both actual and poten- 80 tial) showed smooth diel curves of heating and cooling, attaining their daily maximum late in the afternoon (Fig. 1A). In contrast, E. heatwolei (regardless of sex 60 or reproductive status) achieved high and relatively stable body temperatures as soon as basking opportu- 40 nities were available in the morning (Fig. 1A). Simi- larly high temperatures were often maintained late into the afternoon, but because the proportion of ani- 20 mals continuing to show such high levels declined through the day, so did the mean E. heatwolei temperature (Fig. 1A). Repeated-measures ANOVA on mean 0 values per site (or lizard) per hour, averaged over the 024681012141618202224 entire monitoring period, showed a highly significant B interaction term between hour of day and the type of 35 thermal regime (F69,805 = 28.40, P < 0.0001). (c) The coefficient of variation for temperature was Coefficient of variation 30 higher in the morning than it was in the afternoon, and was higher for E. heatwolei body temperatures 25 than for nests (Fig. 2A). Body temperatures were most 20 variable at around the time of morning emergence, reflecting the rapid rise in mean temperatures at that 15 time of day (Figs 1A, 2A). Variances in body temperature fell rapidly through the later part of the 10 morning, attaining levels similar to those of nest temperatures by the afternoon (Fig. 2A). Repeated- 5 measures ANOVA on mean coefficient of variation (CV) per site (or lizard) per hour, averaged over the 0 entire monitoring period, showed a highly significant 024681012141618202224 interaction term between hour of day and the type of Time of day (h) thermal regime (F69,805 = 15.77, P < 0.0001). (d) Frequency distributions of incubation tempera- Figure 2. Diel shifts in coefficients of variation (CV) for tures were thus unimodal for nest temperatures but hourly temperatures in four types of potential incubation bimodal for E. heatwolei body temperatures (Fig. 3). environment for lizard embryos at a high-elevation site at (e) Day-to-day variation in mean temperatures. To Mount Ginini in the Brindabella Range. The graph shows examine whether differences in these four thermal mean values plus associated standard errors for CV in regimes remained constant through time, I also potential nest sites in the forest (), for actual nests of calculated mean values per site per day for each the oviparous scincid lizard Bassiana duperreyi in open of the replicates. Two-factor repeated-measures clearings ( ), and for body temperatures of male ( ), non- ANOVA revealed a highly significant interaction term reproductive female ( ) and pregnant female ( ) viviparous scincid lizards Eulamprus heatwolei. Each hourly (F12,120 = 10.81, P < 0.0001). Inspection of the data mean value is based on overall mean values for each of the showed a complex pattern whereby daily shifts in replicate sites in each category, averaged over the four days weather affected nests and lizards differently: for of monitoring; for example, each nest contributed only a example, a change in weather conditions from Day 2 to single value for each hour of the day. (A) shows CV values Day 3 reduced mean nest temperatures but increased calculated from raw data, and (B) shows CV calculations mean E. heatwolei body temperatures (Fig. 4). Dis- after all values <16.5 ∞C (the lower thermal limit for crepancies in mean temperatures between the four embryogenesis in B. duperreyi) were replaced with a value types of thermal regimes thus shift with changes in of 16.5 ∞C. Thus, (B) shows the CV in effective mean incu- the weather. bation temperature, allowing for the cessation of develop- (f) Effective mean temperatures for embryonic ment below this threshold level. development will differ from arithmetic mean temperatures because developmental rates do not increase in

80 A Figure 3. Frequency distributions of temperatures exhibited in four types of potential incubation environment for lizard embryos at a high-elevation site at Mount Ginini in 60 the Brindabella Range. The histograms show numbers of records of hourly mean values for (A) potential nest sites 40 in the forest, (B) actual nests of the oviparous scincid lizard Bassiana duperreyi in open clearings, and for body 20 temperatures of (C) male, (D) non-reproductive female and (E) pregnant female viviparous scincid lizards Eulamprus 0 heatwolei. Each animal (or site) provided a single mean 80 B value for each hour of the day, averaged over the four days of monitoring.

20 40

20 18 C)

0 ° 20 C 16

15 14

Mean temperature ( 12 5

0 10 30 D 01234

Number of hourly mean valuess Day no. 25

20 Figure 4. Day-to-day variation in mean temperatures (and associated standard errors) for four types of potential 15 incubation environment for lizard embryos at a high-eleva- 10 tion site at Mount Ginini in the Brindabella Range. The graphs show daily mean values plus associated standard 5 errors for potential nest sites in the forest (), for actual 0 nests of the oviparous scincid lizard Bassiana duperreyi in 40 E open clearings ( ), and for body temperatures of male ( ), non-reproductive female () and pregnant female () viviparous scincid lizards Eulamprus heatwolei. Each daily 30 mean is based on overall means for each of the replicate sites in each category; for example, each nest contributed 20 only a single value per day.

10 irrelevant to embryogenesis (Georges, Limpus & 0 Stoutjesdijk, 1994; Shine & Harlow, 1996). Andrews 0481216202428323640 (2000) suggested a simple way to correct for this effect, Temperature (°C) by substituting the critical thermal threshold for all temperature readings below this value. a simple linear fashion with temperature. Thus, Using this procedure substantially modiﬁes the con- decreases below the minimum level at which embryo- clusions about mean temperatures (section (a) above): genesis ceases (16.5 ∞C in B. duperreyi, Shine & Har- the recalculated mean values were almost identical for low, 1996) have no further effect on development, so actual B. duperreyi nests (19.41 ± 0.21 ∞C), for male that how far temperatures fall below this threshold is E. heatwolei (19.57 ± 0.35 ∞C), non-pregnant female

E. heatwolei (19.50 ± 0.37 ∞C), and pregnant female iparity (compare Fig. 3B vs. Fig. 3C and D) or from E. heatwolei (19.54 ± 0.27 ∞C). Potential nest sites in maternal thermoregulatory adaptations (Fig. 3C and the forest were considerably cooler (17.99 ± 0.26 ∞C) D vs. Fig. 3E), but from maternal selection of nest sites (one-factor ANOVA using mean values per site, in well-exposed areas (Fig. 3A vs. Fig. 3B). Second, the

F4,41 = 7.09, P < 0.0002; Fisher’s PLSD posthoc tests major effect of uterine retention on embryonic thermal showed that potential (forest) nests were cooler than regimes lay in the frequency distribution of tempera- all other regimes at P < 0.001). Figure 1B shows that tures rather than in mean values: nests provided effective incubation temperatures (i.e. ‘corrected’ val- unimodal distributions whereas body temperatures ues) were similar among all four thermal regimes provided bimodal distributions (Fig. 3). overnight, but higher for E. heatwolei than for nests These results have implications for broader issues. during the majority of daylight hours. Repeated- In terms of extending the distributional limits of mon- measures ANOVA on mean ‘corrected’ values per incu- tane reptiles, judicious selection of hot nest sites may bation environment per hour, averaged over the entire be at least as significant as the shift from oviparity to monitoring period, showed a highly significant inter- viviparity. That is, oviposition-site selection behaviour action term between hour of day and the type of ther- by nesting females greatly modifies incubation mal regime (F69,805 = 28.18, P < 0.0001). regimes, and thus the elevational range over which (g) The effective coefficient of variation for incuba- oviparous reptiles can reproduce successfully. Based tion temperature will also be modified by ‘correcting’ on measurements over a wide range of elevations in thermal data as explained above. However, broad diel the Brindabella Range (Shine et al., 2003), the patterns remained similar (albeit, shifted later in the increase in mean incubation temperature due to nest- day: Fig. 2B). E. heatwolei body temperatures were site selection in this study (about 2.5 ∞C) corresponds more variable than were nest temperatures for the to a decrease of about 400 m elevation in mean soil majority of daylight hours, but effective incubation temperatures. Although this study was brief, mean temperatures were essentially constant for most of the nest temperatures that I measured were similar to night in all four types of thermal regime (Fig. 2B). those from long-term (7-year) monitoring of natural Repeated-measures ANOVA on mean CVs of ‘cor- Bassiana nests at this site (Shine, 2002c). rected’ temperatures per site (or lizard) per hour, aver- My data do not show any straightforward thermal aged over the entire monitoring period, showed a benefit to viviparity, at least at intermediate eleva- highly significant interaction term between hour of tions (1615 m a.s.l.). At higher elevations in the day and the type of thermal regime (F69,805 = 20.09, Brindabella Range (e.g. 1720 m a.s.l.), minimum soil P < 0.0001). temperatures fall so low that there are simply no thermally suitable nest sites, and thus viviparity provides the only system which allows maintenance of incuba- DISCUSSION tion temperatures that allow successful incubation Although the duration of monitoring was reduced by a (Andrews, 2000; Shine et al., 2003). Below this level, wildfire, sufficient data were gathered to clarify ther- however, over the range of elevations where oviparous mal regimes for egg incubation available to reptiles at forms coexist with viviparous forms, there may be Mount Ginini. The data were gathered during the little difference in mean incubation temperatures period immediately after oviposition by B. duperreyi, between species of these two reproductive modes and hence during the time when eggs would be (Figs 1 and 2). This situation differs from that usually retained in utero by an oviparous species making the envisaged by proponents of the ‘cold-climate’ model for transition towards viviparity. Thus, we can compare the evolution of viviparity, whereby the persistence of directly the incubation regimes that would be experi- oviparous taxa at relatively high elevations is attrib- enced by: (a) an oviparous species that lays its eggs in uted to a balance between benefits (putatively higher the forest (like N. maccoyi), (b) an oviparous species incubation temperatures for viviparous forms) and that selects warm nest sites in open clearings (like ‘costs’ (reduced maternal mobility due to the physical B. duperreyi), (c) a viviparous species that retains burdening by the litter) (Shine, 1985). Instead, data developing eggs but does not increase basking effort from this study suggest that maternal nest-site selec- due to pregnancy (and thus, will exhibit thermal tion may offer a remarkably effective means of attain- profiles like those of E. heatwolei males and non- ing warm incubation conditions. reproductive females), or (d) a viviparous species in I am reluctant to place too much emphasis on the which females modify their thermoregulatory behav- fact that Bassiana nests exhibited higher arithmetic iour to optimize embryonic development (i.e. pregnant mean temperatures than did gravid Eulamprus, E. heatwolei). The results were counter-intuitive in because restricting the lizards to outdoor arenas may two respects. First, the greatest shift in mean incuba- have prevented them from thermoregulating opti- tion temperatures came not from the evolution of viv- mally. Also, ‘correcting’ these values for mean temper-

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 83, 145–155 152 R. SHINE atures, by taking into account thermal minima for Given the congruence between the results of embryogenesis, eliminated the differences between Andrews (2000) and those of this study, why did ear- nest and body temperatures. The sensitivity of the lier studies (Shine, 1983; Heulin, Osenegg & Lebou- thermal differentials in my study of day-to-day varia- vier, 1991) find that maternal body temperatures tion in weather conditions (Fig. 3) further weakens were substantially higher than nest temperatures? any attempt to make overall statements about which Andrews (2000) noted several possible explanations, conditions are warmer. When I conducted a similar including the location of the previous studies: these study at the same site in the previous summer as part were in laboratory enclosures (Heulin et al., 1991) or of a broader elevational comparison (Shine et al., in areas of sympatry between oviparous and vivipa- 2003), incubation temperatures were about 5 ∞C rous forms (Shine, 1983). Because thermal differen- higher compared with those measured in 2002–2003, tials between nests and females are likely to shift but the thermal differential between pregnant female significantly with elevation (Andrews, 1998, 2000; Eulamprus (mean = 19 ∞C) and Bassiana nests Shine et al., 2003), studies at the extreme elevational (21.5 ∞C) was almost identical to that recorded in this limit of oviparous reproduction may provide the stron- study. Subtle behavioural differences between preg- gest evidence [as in Andrews’s (2000) work as well as nant and non-pregnant lizards may have been appar- in this study]. However, we would expect nest temper- ent if the study had spanned a longer period. A atures to change more rapidly with elevation than previous field study by Schwarzkopf & Shine (1991) would lizard body temperatures (because of the buff- similarly detected no difference in mean temperatures ering effects of behavioural thermoregulation, e.g. see between pregnant E. tympanum and their male coun- Shine et al., 2003). Thus, thermal disparities between terparts, but a more detailed laboratory study of the nests and females should be lower not higher at lower congeneric E. quoyii reported that pregnant females elevations. selected higher and less variable body temperatures Methodological shortcomings in earlier studies may (Borges Landaez, 1999). be a more important reason for the differing conclu- There are few other published reports on thermal sions. For example, my early studies in the Brinda- differentials between reproducing viviparous female bella system calculated a difference of around 7 ∞C in reptiles and nest sites. The most extensive work is mean incubation temperature between eggs in nests that of Andrews (2000), who measured temperatures vs. those retained in utero (Shine, 1983). However, the of nests and female lizards (Sceloporus spp.) at a range evidence used to calculate this figure was indirect, and of elevations in the western USA. She reported results overestimated the thermal differential for three rea- very similar to those presented here. Arithmetic sons: first, I measured maternal body temperatures means for nest vs. female temperatures were virtually only during daylight hours, by capturing active lizards identical in the US system, but diel temperature vari- and measuring their cloacal temperatures. This pro- ation was greater for maternal than for nest temper- cedure may underestimate mean activity tempera- atures. As Andrews (2000) recognized, this means that tures if hotter lizards are more difficult to catch effective incubation temperatures were higher in (Shine, 1983). Equally, however, body temperatures of females than in nests (see Shine & Harlow, 1996 for active animals will exceed mean lizard temperatures further discussion of this phenomenon). However, the if many lizards are inactive during the day (or inactive calculated effect in terms of incubation periods for the all day during inclement weather), and hence are at North American lizards was small except at very great lower body temperatures. Thus, although the 1983 elevations (3 days at 1800 m, 6 days at 2800 m, study concluded that a pregnant female lizard would 17 days at 3200 m). Notably, all of Andrews’s sites average about 26–30 ∞C during the day, in fact the were considerably higher than Mount Ginini, reflect- thermochrons revealed lower and more variable ther- ing the fact that oviparous lizards extend to much mal regimes over this period (Fig. 1). higher elevations in North America than they do in Second, my 1983 calculations assumed that a gravid Australia. Based on these data, Andrews (2000) sug- female lizard could maintain nocturnal temperatures gested that the thermal benefits of initial uterine as high as those in nests. In fact, in this study, the retention of eggs at high elevations would be minor, thermochrons showed that lizards were much colder and probably less significant for overall developmen- at night than were the nearby nest sites (Fig. 1). This tal rates (and thus, hatching times) compared with discrepancy might be due to experimental artefacts, shifting the eggs to a more superficial nest. Thus, she if the enclosures did not contain sufficiently well- argued that the initial adaptation for warmer incuba- insulated hiding-places. Alternatively, this may be a tion in cold climates probably involved maternal nest- real phenomenon: either because females actively site selection rather than uterine retention of eggs. My select cold overnight retreat sites, or because the nest data showed similar patterns as she discovered for sites used by Bassiana are in the middle of the (rela- Sceloporus, and hence support her arguments. tively rare) open clearings and hence are far from the

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 83, 145–155 INCUBATION REGIMES IN COLD-CLIMATE REPTILES 153 usual home ranges of lizards in the surrounding for- This conclusion suggests a subtle shift in interpre- est. Bassiana nests remained hotter late into the tation of the observation that pregnant female reptiles evening than did potential nest sites in the forest of many species thermoregulate at different mean (Fig. 1) but in order to use such a well-insulated noc- temperatures, and with greater precision, than do turnal retreat site, most female lizards would have to non-pregnant animals. This is an extremely wide- travel long distances across open terrain. The risks spread phenomenon in both snakes and lizards (Shine, and energy costs associated with moving so far from 1980; Beuchat, 1986, 1988; Beuchat & Ellner, 1987; the home range every night might well force a vivi- Gier, Wallace & Ingerman, 1989; Charland & Gregory, parous female to accept lower overnight temperatures 1990; Rosen, 1991; Graves & Duvall, 1993; Peterson, than would be accessible in a nest-site exposed to high Gibson & Dorcas, 1993; Tu & Hutchinson, 1994; levels of solar radiation throughout the day. Andrews et al., 1997; Mathies & Andrews, 1997; Dor- Third, my 1983 data on nest temperatures came cas & Peterson, 1998; Rock, Andrews & Cree, 2000; from only four nests, two of which belonged to Rock et al., 2002; Rock & Cree, 2003). Although data N. maccoyi. Subsequent work has shown that Nan- are limited, increased basking effort by gravid females noscincus is unique among the oviparous Brindabella appears to occur in oviparous as well as viviparous scincid taxa in selecting nests in well-shaded (and taxa (Werner, 1990; Blazquez, 1995), and hence should thus cool) sites within the forest, rather than in open be seen as a preadaptation rather than an adaptation areas that attain much higher temperatures (Shine, to viviparity. Many of the species studied in this 1999). Thus, my 1983 study underestimated mean respect have been cool-climate viviparous taxa, in nest temperatures as well as overestimating mean which ambient thermal conditions impose strong chal- body temperatures of females. lenges to behaviourally mediated stenothermy. In More generally, I agree with Andrews (2000) that such environments, increased basking effort by gravid published literature may be overly optimistic about females will result in higher and more stable temper- the efficacy of maternal thermoregulation in vivipa- atures than would otherwise have been the case, but rous reptiles living in cool climates. There can be no females are unlikely to be able to maintain tempera- doubt that behavioural thermoregulation provides a tures as consistently high and stable as they can degree of independence from local weather conditions, attain in laboratory thermal gradients. Thus, direct such that animals can maintain high stable body tem- effects of local weather conditions on embryogenesis peratures even when ambient temperatures are low are potentially important in viviparous as well as and variable (Huey & Slatkin, 1976). Nonetheless, oviparous reptiles. research on reptilian thermoregulation may have This argument predicts that hatchling traits that tended to focus on such ‘success stories’ and pay less are sensitive to incubation temperature (such as num- attention to constraints. Many viviparous reptiles live ber of body segments, and thus ventral scale counts, in places where weather conditions often preclude Vinegar, 1974; Osgood, 1978) should vary through basking, and/or there is little thermal heterogeneity, time and space in concert with local weather condi- and/or the costs of thermoregulation outweigh the tions, in viviparous as well as in oviparous reptiles. In benefits (e.g. Huey, 1974; Huey & Slatkin, 1976; Hertz, agreement with this prediction, a recent study on Huey & Stevenson, 1993; Fitzgerald, Shine & Lemck- Vipera aspis in France showed that meristic traits of ert, 2003). Especially in cold high-elevation sites, hatchling vipers near the northern (cold-climate) edge maintenance of high stable temperatures during the of the species’s distribution varied among years brief periods of time when behavioural thermoregula- depending upon ambient temperatures (Lourdais tion is effective will have relatively little impact on et al., 2004). Climate-correlated geographical clines in mean body temperatures when calculated over a 24-h ventral scale counts in several squamate species period. Hence, local weather conditions may constrain (Klauber, 1941) may reflect the same thermal link body temperatures substantially in many cold-climate between maternal and ambient temperatures. In reptile species, with the result that maternal thermal short, maternal thermoregulation may weaken the regimes may be more similar to nest temperatures constraints imposed by ambient conditions, but cannot than would be predicted from the paradigm (adopted break that link entirely. uncritically in my 1983 study) that heliothermic squa- We will need many more studies, over longer time mates can maintain high, stable body temperatures periods and a range of sites, to evaluate the generality for much of the day. Indeed, body temperatures of of results from this study. Meanwhile, we are left with E. heatwolei (including pregnant females) were more a result that runs counter to conventional wisdom: at rather than less variable compared with temperatures intermediate elevations, maternal nest-site selection in potential or actual nest sites, regardless of whether may be at least as important as is the evolution of viv- the calculations were corrected for embryonic develop- iparity, or adaptations of maternal thermoregulatory mental rates (Fig. 2). behaviour in viviparous species, in allowing reptiles to

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 83, 145–155 154 R. SHINE reproduce in severely cold climates. At the least, it is Fitzgerald M, Shine R, Lemckert F. 2003. A reluctant helio- dangerous to assume that thermal modifications asso- therm: thermal ecology of the arboreal snake Hoplocephalus ciated with viviparity have had more effect on embry- stephensii (Elapidae) in dense forest. Journal of Thermal onic environments than has maternal nest-site Biology 28: 515–524. selection in oviparous taxa, or that embryos of a vivi- Georges A, Limpus CJ, Stoutjesdijk R. 1994. Hatchling sex parous species inevitably will be kept at higher tem- in the marine turtle Caretta caretta is determined by propor- peratures through incubation than will the embryos of tion of development at a temperature, not daily duration of a sympatric oviparous taxon. exposure. Journal of Experimental Zoology 270: 432–444. Gibbons JW, Scott DE, Ryan TJ, Travis J, Buhlmann KA, Tuberville TD, Metts BS, Greene JL, Mills T, Leiden Y, ACKNOWLEDGEMENTS Poppy S, Winnie CT. 2000. The global decline of reptiles, I thank Melanie Elphick and Elizabeth Barrott for deja vu amphibians. Bioscience 50: 653–666. field assistance, David Elphick for accommodation and Gier PJ, Wallace RL, Ingerman RL. 1989. Influence of preg- hospitality, and Anna Farham and Steve Taylor (ACT nancy on behavioral thermoregulation in the Northern National Parks Service) for rescuing the lizards from a Pacific rattlesnake. Crotalus viridis oreganus. Journal of fiery death. This study was funded by the Australian Experimental Biology 145: 465–469. Research Council. Graves BM, Duvall D. 1993. Reproduction, rookery use, and thermoregulation in free-ranging, pregnant Crotalus v. viri- REFERENCES dis. Journal of Herpetology 27: 33–41. Greer AE. 1989. The biology and evolution of Australian liz- Andrews RM. 1998. Geographic variation in field body tem- ards. Sydney: Surrey Beatty and Sons. perature of Sceloporus lizards. Journal of Thermal Biology Greer AE. 1997. The biology and evolution of Australian 23: 329–334. snakes. Sydney: Surrey Beatty and Sons. Andrews RM. 2000. Evolution of viviparity in squamate rep- Hertz PE, Huey RB, Stevenson RD. 1993. Evaluating tem- tiles (Sceloporus spp.): a variant of the cold-climate model. perature regulation by field-active ectotherms: the fallacy of Journal of Zoology 250: 243–253. the inappropriate question. American Naturalist 142: 796– Andrews RM, DelaCruz FRM, SantaCruz MV. 1997. Body 818. temperatures of female Sceloporus grammicus: thermal Heulin B, Osenegg K, Lebouvier M. 1991. Timing of embry- stress or impaired mobility? Copeia 1997: 108–115. onic development and birth dates in oviparous and vivipa- Beuchat CA. 1986. Reproductive influences on the thermoreg- rous strains of Lacerta vivipara: testing the predictions of an ulatory behaviour of a live-bearing lizard. Copeia 1986: 971– evolutionary hypothesis. Acta Oecologia 12: 517–528. 979. Huey RB. 1974. Behavioral thermoregulation in lizards: Beuchat CA. 1988. Temperature effects during gestation in a importance of associated costs. Science 184: 1001–1003. viviparous lizard. Journal of Thermal Biology 13: 135–142. Huey R, Slatkin M. 1976. Costs and benefits of lizard ther- Beuchat CA, Ellner S. 1987. A quantitative test of life his- moregulation. Quarterly Review of Biology 51: 363–384. tory theory: thermoregulation by a viviparous lizard. Ecolog- Hughes L, Cawsey EM, Westoby M. 1996a. Climatic range ical Monographs 57: 45–60. sizes of Eucalyptus species in relation to future climate Blackburn DG. 1982. Evolutionary origins of viviparity in the change. Global Ecology and Biogeography 5: 23–29. Reptilia. I. Sauria. Amphibia-Reptilia 3: 185–205. Hughes L, Cawsey EM, Westoby M. 1996b. Geographic and Blackburn DG. 1985. Evolutionary origins of viviparity in the climatic range sizes of Australian eucalypts and a test of Reptilia. II. Serpentes, Amphisbaenia, and Icthyosauria. Rapoport’s rule. Global Ecology and Biogeography 5: 128– Amphibia-Reptilia 6: 259–291. 142. Blazquez MC. 1995. Body temperature, activity pattern and Klauber LM. 1941. The correlation between scalation and life movements by gravid and non-gravid females of Malpolon zones in San Diego County snakes. Bulletin of the Zoological monspessulanus. Journal of Herpetology 29: 264–266. Society of San Diego 17: 73–79. Borges Landaez PA. 1999. Maternal thermoregulation and Langkilde T, O’Connor D, Shine R. 2003. Shelter-site use its consequences for offspring fitness in the Australian east- by five species of montane scincid lizards in southeastern ern water skink (Eulamprus quoyii). PhD Thesis, Biological Australia. Australian Journal of Zoology 51: 175–186. Sciences, University of Sydney, Sydney, NSW, Australia. Lourdais O, Shine R, Bonnet X, Guillon M, Naulleau G. Burger J. 1989. Incubation temperature has long-term effects 2004. Climate affects offspring phenotypes in a viviparous on behaviour of young pine snakes (Pituophis melanoleucus). snake. Oikos 104: 551–560. Behavioral Ecology and Sociobiology 24: 201–207. Mathies T, Andrews RM. 1997. Influence of pregnancy on Charland MB, Gregory PT. 1990. The influence of female the thermal biology of the lizard, Sceloporus jarrovi: why do reproductive status on thermoregulation in a viviparous pregnant females exhibit low body temperatures? Functional snake, Crotalus viridis. Copeia 1990: 1089–1098. Ecology 11: 498–507. Dorcas ME, Peterson CR. 1998. Daily body temperature Olsson M, Shine R. 1997. The seasonal timing of oviposition variation in free-ranging rubber boas. Herpetologica 54: 88– in sand lizards (Lacerta agilis): why earlier clutches are bet- 103. ter. Journal of Evolutionary Biology 10: 369–381.

Osgood DW. 1978. Effects of temperature on the development Shine R. 1987. Reproductive mode may determine geographic of meristic characters in Natrix fasciata. Copeia 1978: 33–47. distributions in Australian venomous snakes (Pseudechis, Pengilley R. 1972. Systematic relationships and ecology of Elapidae). Oecologia 71: 608–612. some lygosomine lizards from southeastern Australia. PhD Shine R. 1999. Egg-laying reptiles in cold climates: determi- Thesis, Australian National University. nants and consequences of nest temperatures in montane liz- Peterson CR, Gibson AR, Dorcas ME. 1993. Snake thermal ards. Journal of Evolutionary Biology 12: 918–926. ecology: the causes and consequences of body-temperature Shine R. 2002a. Reconstructing an adaptationist scenario: variation. In: Seigel RA, Collins JT, eds. Snakes: ecology and what selective forces favor the evolution of viviparity in mon- behavior. New York: McGraw-Hill, 241–314. tane reptiles? American Naturalist 160: 582–593. Qualls CP. 1997. The effects of reproductive mode and climate Shine R. 2002b. Eggs in autumn: responses to declining incu- on reproductive success in the Australian lizard, Lerista bou- bation temperatures by the eggs of montane lizards. Biolog- gainvillii. Journal of Herpetology 31: 60–65. ical Journal of the Linnean Society 76: 71–77. Qualls R, Shine R, Donnellan S, Hutchinson M. 1996. The Shine R. 2002c. An empirical test of the ‘predictability’ evolution of viviparity within the Australian scincid lizard hypothesis for the evolution of viviparity in reptiles. Journal Lerista bougainvillii. Journal of Zoology (London) 237: 13– of Evolutionary Biology 15: 553–560. 26. Shine R. 2004. Seasonal shifts in nest temperature can modify Robert KA, Thompson MB. 2003. Reconstructing Thermo- the phenotypes of hatchling lizards, regardless of overall chron iButtons to reduce size and weight as a new technique mean incubation temperature. Functional Ecology 18: 43– in the study of small animal thermal biology. Herpetological 49. Review 34: 130–132. Shine R, Barrott EG, Elphick MJ. 2002a. Some like it hot: Robertson P. 1981. Comparative reproductive ecology of two effects of forest clearing on nest temperatures of montane south-eastern Australian skinks. In: Banks CB, Martin AA, reptiles. Ecology 83: 2808–2815. eds. Proceedings of the Melbourne herpetological symposium. Shine R, Elphick MJ. 2001. The effect of short-term weather Melbourne: Zoological Board of Victoria, 25–37. fluctuations on temperatures inside lizard nests, and on the Rock J, Andrews RM, Cree A. 2000. Effects of reproductive phenotypic traits of hatchling lizards. Biological Journal of condition, season, and site on selected temperatures of a the Linnean Society 72: 555–565. viviparous gecko. Physiological and Biochemical Zoology 73: Shine R, Elphick M, Barrott EG. 2003. Sunny side up: 344–355. lethally high, not low, temperatures may prevent oviparous Rock J, Cree A. 2003. Intraspecific variation in the effect of reptiles from reproducing at high elevations. Biological Jour- temperature on pregnancy in the viviparous gecko Hoplodac- nal of the Linnean Society 78: 325–334. tylus maculatus. Herpetologica 59: 8–22. Shine R, Elphick MJ, Donnellan S. 2002b. Co-occurrence of Rock J, Cree A, Andrews RM. 2002. The effect of multiple, supposedly incompatible modes of sex determina- reproductive condition on thermoregulation in a viviparous tion in a lizard population. Ecology Letters 5: 486–489. gecko from a cool climate. Journal of Thermal Biology 27: Shine P, Elphick MJ, Harlow PS. 1997. The influence of 17–27. natural incubation environments on the phenotypic traits of Rosen PC. 1991. Comparative field study of thermal prefer- hatchling lizards. Ecology 78: 2559–2568. enda in garter snakes (Thamnophis). Journal of Herpetology Shine P, Harlow PS. 1996. Maternal manipulation of off- 25: 301–312. spring phenotypes via nest-site selection in an oviparous liz- SAS Institute. 1998. Statview 5. Cary, Indiana: SAS Institute. ard. Ecology 77: 1808–1817. Schwarzkopf L, Shine R. 1991. Thermal biology of reproduc- Tu MC, Hutchinson VH. 1994. Influence of pregnancy on tion in viviparous skinks, Eulamprus tympanum: why do thermoregulation of water snakes (Nerodia rhombifera). gravid females bask more? Oecologia 88: 562–569. Journal of Thermal Biology 19: 255–259. Sexton OJ, Claypool L. 1978. Nest sites of a northern pop- Vinegar A. 1974. Evolutionary implications of temperature ulation of an oviparous snake, Opheodrys vernalis (Ser- induced anomalies of development in snake embryos. Herpe- pentes, Colubridae). Journal of Natural History 12: 365–370. tologica 30: 72–74. Shine R. 1980. ‘Costs’ of reproduction in reptiles. Oecologia Warner DA, Andrews RM. 2002. Laboratory and field exper- 46: 92–100. iments identify sources of variation in phenotypes and sur- Shine R. 1983. Reptilian viviparity in cold climates: testing vival of hatchling lizards. Biological Journal of the Linnean the assumptions of an evolutionary hypothesis. Oecologia 57: Society 76: 105–124. 397–405. Werner YL. 1990. Do gravid females of oviparous gekkonid Shine R. 1985. The evolution of viviparity in reptiles: an eco- lizards maintain elevated body temperatures? Hemidactylus logical analysis. In: Gans C, Billett F, eds. Biology of the Rep- frenatus and Lepidodactylus lugubris in Oahu. Amphibia- tilia, Vol. 15. New York: John Wiley & Sons, 605–694. Reptilia 11: 200–204.