BzoEogical Journal of the Linnean Society (2001), 72: 555565. With 5 figures doi:10.1006/bij1.2000.0516, available online at http;//www.idealibrary.com on I Of @ The effect of shortterm weather fluctuations on temperatures inside lizard nests, and on the phenotypic traits of hatchling lizards RICHARD SHINE* and MELANIE J. ELPHICK Biological Sciences AO8, University of Sydney, NSW 2006, Australia Received 21 May 2000; accepted for publication 20 December 2000 Can shortterm stochastic variation in local weather conditions modify the thermal conditions inside lizard nests, and thus (potentially) the developmental rates, hatching success, and phenotypic traits of hatchlings from these nests? This hypothesis requires that (i) natural nests are poorly buffered thermally, such that ambient regimes affect temperatures inside the nest, and (ii) shortterm thermal variations modify attributes of the offspring. Field data on natural nests of the subalpine skink Bassiana duperreyi confirm the existence of this first effect, and laboratory experiments substantiate the latter. Exposure to warmerthanusual temperatures for 2 weeks during the 9 to 16week incubation period doubled hatching success, and significantly modified hatchling phenotypes (hatching dates, offspring size and locomotor performance). The proportion of development completed prior to this exposure influenced the degree of response. Exposure to a brief window of higherthanusual temperatures soon after oviposition had more effect on hatching time, egg survival and hatchling phenotypes than if the exposure occurred later in development. Thus, minor variations in weather conditions during incubation may have substantial effects on reptile populations. 02001 The Linnean Society of London ADDITIONAL KEYWORDS: Bassiana duperreyi incubation phenotypic plasticity reptile. INTRODUCTION The thermal environment is an important influence on the daytoday lives of many kinds of animals, especially ectotherms. For example, many reptiles use complex behavioural and physiological thermoregulation to maintain their body temperatures within acceptable limits (e.g. Huey & Slatkin, 1976 Avery, 1982; Avery, Bedford & Newcombe, 1982; Peterson, Gibson & Dorcas, 1993). But what of other lifehistory stages, such as the egg? A reptilian embryo cannot control its own temperature, and hence may be subjected to substantial thermal variation. Such variation may have important consequences, because thermal conditions during reptilian incubation can affect not only hatching success, but also many phenotypic traits of the hatchling such as its sex, size, shape, colour, behaviour and locomotor performance (e.g. Bull, 1980; Burger, Zappalorti & Gochfeld, 1987; Burger & Zappalorti, 1988; Burger, 1989). Hence, the ways in which * Corresponding author. Email: rics@bio.usyd.edu.au 00244066/01/040555 + 11 $35.00/0 555 thermal variation affect the biology of reptilian embryogenesis deserve further study. Previous studies in this field have focused on the effects of a mother s nestsite choice on the phenotypes of her offspring (e.g. above references, and Viets et al., 1993; Shine & Harlow, 1996; Shine et at., 1996). Thus, we have considerable data to show that embryos developing under one thermal regime throughout incubation may develop in ways that differ from those that experience a different thermal regime. However, virtually all studies in this field have incubated each individual egg under the same conditions throughout development. Differences among eggs in thermal regimes have thus been applied over the entire developmental period. This experimental design grossly oversimplifies the conditions experienced by developing eggs in natural nests: thermal conditions may vary not only among nests, but also from daytoday or weektoweek within any given nest (Packard, Tracy & Roth, 1977; Packard & Packard, 1988; Shine & Harlow, 1996). Although temporal variation in incubation conditions is likely to be common, it has attracted little 0 2001 The Linnean Society of London
556 R. SHINE and M. J. ELPHICK scientific attention. Particularly for relatively shallow nests, daytoday or weektoweek variation in weather conditions (temperature, solar radiation, precipitation) may plausibly have substantial effects on physical conditions inside the nest. Is this true? And even if it is, can such stochastic variation engender significant variation in hatchling phenotypes? If so, what kinds of traits are affected, and to what degree? Does the embryo s response to such shortterm perturbations change throughout the course of development? For example, we might expect that offspring become progressively less sensitive as incubation proceeds, because many of the developmental changes have already occurred. We gathered data on scincid lizards to evaluate two questions: (i) do temperatures inside lizard nests vary appreciably in response to shortterm weather conditions? and (ii) does a relatively brief exposure of eggs to higher temperatures influence hatching success, incubation period or hatchling phenotypes? MATERIAL AND METHODS STUDY SPECIES The threelined skink, Bassiana duperreyi (Leiolopisma trilineatum in older literature) is an oviparous, mediumsized lizard (to 80 mm snoutvent length) that is widely distributed in subalpine southeastern Australia (Cogger, 1992). Females lay a single clutch of three to nine eggs in summer (Pengilley, 1972; Shine, 1980, 198313; Shine & Harlow, 1996). MEASURING TEMPERATURES IN NATURAL NESTS To examine variation in thermal regimes over the course of incubation, we placed portable dataloggers (HoboTemp, Onset Computer Corp., Pocasset, Mass.) in natural nests of Bassiana duperreyi shortly after eggs were laid in early December each year. Data were obtained for the entire incubation period for 60 nests over a period of 5 years (1994/95 to 1998/99). These nests were distributed among three sites in the Brindabella Ranges, approximately 40 km west of Canberra in the Australian Capital Territory. The sites were as follows: Coree Flats (elevation 1050 m), Piccadilly Circus (1246 m) and Ginini Flats (1600 m). The first two sites are 9km apart, whereas Ginini Flats is further away (24km south of Piccadilly Circus). The probes were placed among freshlylaid eggs, such that eggs were both above and below the probe. In many cases, females deposited additional eggs around the probes after they were placed in position. These dataloggers were set to record temperatures every 60 min for the entire incubation period. PHENOTYPIC RESPONSES OF HATCHLINGS TO TRANSIENT THERMAL VARIATION In December 1997, we collected recentlylaid Bassiana eggs from natural nests at the three sites described above (13 clutches from Coree Flats, 46 from Ginini Flats and 17 from Piccadilly Circus). Eggs were individually weighed to the nearest 0.001 g on a Sartorius toploading balance. They were then transferred to individually labelled 64 ml glass jars containing vermiculite (water potential = 200 Wa), and sealed with plastic cling wrap to prevent moisture loss during incubation. The eggs were then randomly assigned to incubation treatments. Clayson 10step programmable incubators were set to create daily cycles of heating and cooling, to mimic thermal regimes of Bassiana nests in the field (recorded with temperature loggers, as above: Shine & Harlow, 1996). Three such incubators were used. One of them was set at conditions typical of those experienced in relatively cool natural nests ( cool incubator, diel cycle 17 f 5 C). Another was set at temperatures typical of warm natural nests ( warm incubator, diel cycle 22f5 C), and the third was set even hotter ( hot incubator, diel cycle 27 5OC). This third treatment is hotter than mean conditions experienced in any natural nests, but was included to mimic circumstances that may occur transitorily during very hot weather. All eggs were initially placed in the cool incubator, and some of them (n = 133) remained there throughout incubation. However, most of the eggs (n = 320) were transferred to one of the other incubators for a 2week period at some point of development. These eggs were then returned to the cool incubator to complete development. The various treatment groups differed in: (i) the thermal regime experienced during the 2week thermal window (either the warm or hot incubator); and (ii) the time in development at which that transfer occurred. One group of 40 eggs (with approximately equal numbers from each population) was transferred at the beginning of incubation; another set 1 week later; another set 2 weeks later, and so forth. Thus, the first group spent weeks 1 and 2 at higher temperatures; the second group spent weeks 2 and 3 at these temperatures; the third group spent weeks 3 and 4 at these temperatures, and so forth. Because of financial constraints, our experimental design did not incorporate replication at the level of incubators within thermal treatments. Thus, any treatment effects are potentially due to some characteristic of the incubator used, rather than temperature per se. The same difficulty applies to most experimental work in this field. In the case of our study, indirect evidence suggests that temperature is indeed the most important causal influence on our results: (1) Relationships between incubation temperature
and hatchling phenotype are similar in studies using different incubators, and similar in the field (in natural nests) to those reported in laboratory studies (Shine, Elphick & Harlow, 1997); No factor other than temperature has been shown to influence developmental rates of squamate eggs (e.g. Packard & Packard, 1988); and The direction and degree of treatment effects on incubation periods are all consistent with predictions based on previous reports of the thermal dependence of developmental rates in this species (Shine 87, Harlow, 1996; Shine et al., 1997). HATCHLING MORPHOLOGY AND HUSBANDRY Incubators were checked daily, and all hatchlings were weighed ( & 0.001 g), measured (snoutvent length [ = SVL] and tail length), sexed and individually toeclipped (see Harlow, 1996 and Shine et al., 1997 for sexing techniques). Hatchlings from all treatment groups were maintained under identical conditions once they left the incubators. Five lizards were housed per 22 x 13 x 7 cm plastic container. These containers were kept in a controlled temperature room at 18 f 1 C. However, all containers were placed on heating racks such that one end of the cage was situated over a heating cable. This arrangement enabled the hatchlings to thennoregulate over a gradient ranging from 23 to 38 C for 9 h each day (09001800 h). This range brackets the body temperatures selected by active B. dupermyi in the field (Shine, 198313). The room lights were programmed to create a cycle of 10 h lighk14 h dark. Hatchlings were fed a mixture of mealworms (Tenebrio larvae) and crickets twice weekly. Water was provided ad libitum. Hatchlings were reweighed and measured at 1 and 2 weeks of age before being released at their site of collection. LOCOMOTOR PERFORMANCE OF HATCHLINGS The locomotor ability of Bassiana duperreyi offspring was assessed at 1 and 2 weeks of age. Running speeds were measured in a temperaturecontrolled room at 25 C (k 1 C). Hatchlings were transported to the trial room in individual holding jars and left to acclimate undisturbed for at least 30min prior to running. Lizards were then transferred to the holding area of the raceway before being released and allowed to run the lm distance. An artist s paintbrush was used to keep the lizards running. Each hatchling was raced three times, with at least 15 min rest between successive trials. Running speeds (m/s) were determined with the aid of an infrared timing device. Photocells located at 25cm intervals along the runway allowed us to record the cumulative time taken for lizards to cover each 25cm interval. We calculated mean running speed LIZARD NEST TEMPERATURES 557 over 1 m, and mean, burst speed (defined as the fastest speed recorded over a 25 cm distance), for each lizard. STATISTICAL ANALYSES Because most nests of Bassiana are communal (e.g. Pengilley, 1972), we could not always identify individual clutches with confidence. Thus, we did not use clutch number as a factor in our analyses. However, we included source population as a factor, to allow for the possibility of differences among eggs from the three collection sites. Because some of the variables that we measured are not independent of each other, we used residual scores and multivariate analysis of variance (MANOVA) to analyse data for subsets of these traits. For example, three traits associated with body size (mass, snoutvent length and tail length) were measured at each age (hatching, 1 week, 2 weeks). These measures were highly correlated with each other. Thus, our analyses used the following three variables to represent hatchling morphology: SVL, and the residual scores from the linear regressions of mass against SVL and tail length against SVL. This procedure provided three measures that were not intercorrelated. Because they were all based on the same individuals, however, we used MANOVA rather than ANOVA to analyse morphological data for hatchlings of each age (1 and 2 weeks). Similarly, running speeds over two distances (25 cm and 1 m) were combined for MANOVA analysis. Age groups were kept separate in these analyses because we were interested in whether or not the effects of incubation conditions persisted for some time after hatching. Analyses of the responses of hatchlings to transient thermal variation were based on comparisons among three groups of eggs from each of the populations. One group was kept in the cool incubator throughout incubation, whereas the other groups experienced a brief (2week) period in either the warm or hot incubator. These two latter sets of eggs were further divided into several treatment groups that differed in the time during incubation when they were moved to a highertemperature treatment. We conducted two types of comparisons on these data: First, we compared the three main groups with each other. This enabled us to determine if a brief period in a warmer incubator affected hatching success and phenotypic traits of the hatchlings. We could also see if transfer to a hot rather than warm incubator at this time made any difference to these traits. Then, we deleted data for the groups that were kept in the cool incubator for the entire period, and looked only at the eggs that were exposed to a thermal window. Within this group, we could judge whether the timing of exposure, as well
558 R. SHINE and M. J. ELPHICK 35 r IA 18 Dec 95 11 Jan 96 5 Feb 96 40 ib 35 9 30 v 0) 5 25 42 20 g E 15 101 I 18 be, 97 14 Jan 98 30 Jan 98 Figure 1. Temperatures inside two natural nests of the scincid lizard Bassiana duperreyi in the Brindabella Mountains of southeastern Australia. These records were selected to display shortterm variation in mean and maximum nest temperatures due to shifts in weather conditions, as the temperature during that period, affected hatching success or hatchling phenotypes. RESULTS TEMPORAL VARIATION IN THERMAL REGIMES WITHIN NATURAL NESTS Nest temperatures varied considerably, over a variety of timescales. For example, mean nest temperatures varied significantly among the three sites we used (F21H=7.71, RO.01) and among the three years for which we have data on all three sites (F2,48= 12.66, P<O.OOl). Die1 variation was typically high, as demonstrated in previous studies on these populations (e.g. Shine & Harlow, 1996; Shine et al., 1997, and see Fig. 1). However, the magnitude of diel thermal variation varied substantially: periods of low solar insolation produced relatively constant temperatures for days at a time (e.g. see Fig. 1). Mean nest temperatures varied over similar timescales. Commonly, both the mean and variance of temperatures within a single nest often shifted substantially from one day to the next (e.g. see Fig. 1). HOW REALISTIC ARE OUR INCUBATION TREATMENTS? The field data (above) show that Bassiana nest temperatures reflect changes in ambient thermal conditions. What is the average magnitude and timing of these shifts, compared with our experimental treatments during incubation? That is, do eggs in natural nests experience thermal shifts of approximately this magnitude? Temporal shifts in the mean and diel range of nest temperatures were examined over two time periods: from daytoday and from weektoweek. Both the mean and the range of nest temperatures often changed significantly over both of these time periods (Fig. 2). Shifts in mean temperature of 3 or 4 C were commonly observed from one week to the next, but successive daily means rarely differed by such a large amount (Fig. 2). Shifts in thermal variance also occurred at each timescale we investigated (Fig. 2), and importantly, these two traits shift in parallel. Minimum daily temperatures remain fairly constant, and the shifts in mean temperature are generated by changes in maximum temperature (and thus, diel range: see Fig. 1). For example, changes in mean nest temperatures were closely associated with concurrent shifts in thermal ranges, regardless of whether the comparisons involved shifts from daytoday (n = 536, r=0.59, P<O.OOOl) or from weektoweek (n=405, r= 0.49, P<O.OOOl). This result is directly relevant to embryonic development in Bassiana, because an increase in diel thermal range has the same effect as an increase in mean temperature (Shine & Harlow, 1996). For example, an increase in the diel thermal range of nest temperatures by 12 C (i.e. range of 19.5 rather than 75 C) had the same effect on incubation period as an increase of 5 C in mean temperature (Shine & Harlow, 1996). Hence, temporal changes in effective natural nest temperatures (as experienced by developing Bassiana embryos) will be greater than would be inferred from shifts in mean temperature, because of the simultaneous shifts in thermal variance. Thus, many eggs in cool natural nests will be exposed to considerably warmerthanusual temperatures for at least 2 weeks (as in our laboratory trials), although these warm days may be divided into several short periods rather than one long one (Fig. 2). We conclude that at least one of our thermal treatments (the warm treatment) may be close to biological reality. DOES BRIEF EXPOSURE TO HIGHER TEMPERATURES AFFECT HATCHING SUCCESS? The eggs were classified into nine groups, based on the three sites where they were collected and the three alternative thermal treatments during incubation. These
LIZARD NEST TEMPERATURES 559 140 : m 120 E 100 cr a Q 80 E, 60 40 20 n Y 70 60 76543210 1 2 3 4 5 6 7 Daily change in mean temperature ( C) 16 12 8 4 0 4 8 12 16 20 14 10 6 2 2 6 10 14 18 22 Daily change in thermal range ( C) 70 ID 40 30 20 10 4321 0 1 2 3 4 5 6 7 8 9 10 Weekly change in mean temperature ( C) n 9 7 5 3 1 1 3 5 7 9 11 13 15 17 19 8 6 4 2 0 2 4 6 8 10 12 14 16 18 20 Weekly change in thermal range ( C) Figure 2. Temporal variation in thermal regimes within natural nests of the scincid lizard Bassiana duperreyi. These graphs provide data on two different thermal attributesmean nest temperature and the diel range in nest temperatureover two different temporal scales (A&B, daytoday; C&D, weektoweek). In each case, the graphs show frequency distributions for the magnitude of temporal shifts in these variables. The numbers were calculated by subtracting data (on either mean or diel thermal range) for 1 day or week from the corresponding data for the preceding day or week. Analyses of daytoday and weektoweek variation are based on a randomlyselected subset of nine nests monitored over 4 years. thermal groups comprised (i) eggs that had a window of hot conditions; (ii) those with a window of warm conditions; and (iii) those which remained in the cool incubator throughout development. This latter group can be thought of as having a window of cool conditions. We used logistic regression (with hatch or not as the dependent variable, and site and window temperature [hot, warm or cool] as the factors) to evaluate determinants of hatching success. Loglikelihood ratio tests from this regression showed that hatching success varied among sites (xz= 12.21,2 df, R0.016). More importantly, hatching success was significantly affected by the thermal window experienced by the eggs (x2 = 126.72,2 df, P<O.OOOl). This latter effect was due to a much lower hatching success in eggs maintained throughout incubation at cool temperatures (63 of 133 eggs, = 47%) compared to eggs exposed to a 2week window of either warm or hot conditions (respectively, 154 of 160, = 96%; and 153 of 160 eggs, = 96%). That is, hatching success was doubled by a 2week exposure to only slightly warmer conditions. DOES THE TIMING OF EXPOSURE TO HIGHER TEMPERATURES AFFECT HATCHING SUCCESS? Brief exposure to higher temperatures was associated with substantially increased hatching success (see above). Does the effect also depend upon the stage of development at which this exposure occurred? Again, we used logistic regression to disentangle effects of site, window temperature and time of exposure on hatching success. Unlike the previous analysis, this one was based only on eggs exposedto a highertemperature window (because we cannot sensibly talk of the timing of a cool window, if eggs are kept at the same temperatures throughout development). As before, the regression
~ 560 R. SHINE and M. J. ELPHICK Table 1. Effects of a 2week exposure to higher temperatures during embryogenesis, on incubation periods and hatchling phenotypes of the scincid lizard Bassiana duperreyi. The table provides results from twofactor ANOVA (for incubation period) or MANOVA (for other traits). In both cases, the factors are site of collection and the temperature treatment during the thermal window. There are three such thermal treatments: no window (i.e. kept under cool conditions, = 17 k 5 C), warm window (22 k 5 C) or hot window (27 5 C). Significance levels for MANOVA are based on Wilks Lambda. Sample sizes (and thus, degrees of freedom) decrease with age because of mortality and tail loss. Statistically significant results (P<0.05) are shown in boldface font Trait Effect of Incubation period (days) BODY SIZE AND SHAPE: at hatching at 1 week of age at 2 weeks of age RUNNING SPEED (WS): at 1 week of age at 2 weeks of age Site Window temperature Interaction F2,414 = 2 8 ~ ~ F6,818= 12.19 F6,6~=6.10 F6,576 = 4.03 P=O.O006 F4630 = 2.60 P = 0.035 F4,wz = 2.72 P=0.029 showed that hatching success varied among sites (x2 = 6.86, 2 df, P<0.04), but did not differ between eggs exposed to warm vs. hot thermal windows (xz=0.15, ldf, P=O.90). Importantly, however, the time of exposure to higher temperatures modified hatching success (using week of exposure as a continuous variable, x2 = 4.54, 1 df, B0.035). This effect reflects a slightly lower hatching success for eggs exposed to the thermal window in the latter part of embryogenesis. Another way to quantify this effect is withcontingencytable analysis, comparing the relative numbers of eggs hatching vs. failing to hatch after exposure to higher temperatures in the first (14 weeks postlaying) vs. second (58 weeks postlaying) halves of our treatment week groups (x2=5.13, 1 df, P<0.03). DOES BRIEF EXPOSURE TO HIGHER TEMPERATURES AFFECT HATCHLING PHENOTYPES? We used twofactor ANOVA and MANOVA (with site and window temperature [cold, warm, hot] as the factors) to examine whether brief exposure to higher temperatures affected hatchling phenotypes as well as hatching success. The clear result was that the thermal window modified incubation periods, hatchling morphology and running speeds (Table 1). Interpretation is simplified by the lack of significant interactions between factors for traits other than incubation period (Table 1). Visual inspection and posthoc tests reveal F2.414 = 504.3 Feals = 20.45 F6.626 = 14.03 F6,576= 13.02 F4,630= 18.52 F4,,, = 12.76 P= o.ooo1 an overall pattern whereby exposure to a hot window accelerated development (reduced incubation periods) and affected phenotypes more than did exposure to a warm window. Thus, the warm treatment group were intermediate between the hot and cold groups for mean values of all traits. Interestingly, the warm and hot groups were generally more similar to each other than either was to the cold (=no thermal window) group, indicating that even a slight increase in temperature over a 2week period was enough to modify all of the traits that we measured. The magnitude of these effects was substantial, especially for incubation period. The mean duration of incubation was 115 days for eggs kept in the cool incubator throughout development, compared to only 92 days for the eggs kept warm for 2 weeks, and 62 days for the eggs kept hot over the 2week period. That is, a 14day exposure to 27 C rather than 17 C reduced the overall incubation period by 53 days. This result suggests that even a few hot days during incubation can significantly accelerate the hatching date under field conditions. DOES THE TIMING OF EXPOSURE TO HIGHER TEMPERATURES AFFECT HATCHLING PHENOTYPES? Given that exposure to higher temperatures affects the phenotypes of hatchling lizards (Table l), does the magnitude of this effect depend upon the time in
LIZARD NEST TEMPERATURES 561 Table 2. Effects of a brief exposure to higher temperatures during embryogenesis, on incubation periods and hatchling phenotypes of the scincid lizard Bassiana duperreyi. The two thermal treatments involved 2 weeks at either 22 f 5 C or 27 +5 C. For the rest of incubation the eggs were kept at 17+5 C. The Table provides results from threefactor ANOVA (for incubation period) or MANOVA (for other traits). In both cases, the factors are site of collection, the temperature treatment during the thermal window, and the week into incubation when the thermal treatment commenced. For interaction terms, only statistically significant results are shown; t = window temperature, w = week of exposure; s = site. Significance levels for MANOVA are based on Wilks Lambda. Sample sizes decrease with age because of mortality and tail loss. Statistically significant results (P<0.05) are shown in boldface font Trait Effect of: Site Window Week of exposure Interaction temperature Incubation period (days) FZ,, = 73.54 FIB1 = 1188.8 F7S1 = 10.19 TxRTxS; WxS;WxSxT BODY SIZE AND SHAPE: at hatching F,,,4 = 12.76 F3*1= 29.25 Fz1,7== 2.78 wxs P=O.ool at 1 week of age F G ~ 12.61 = Fsm = 22.87 FzI,m= 3.45 wxs P= o.ooo1 at 2 weeks of age F6,434 = 8.69 Fs$i,= 13.74 F21,W = 2.19 none P= o.oo06 P=0.0017 RUNNING SPEED m): at 1 week of age F4.M = 2.47 F4W = 18.27 F14,484= 1.61 WxS;WxT P=O.044 P= 0.07 at 2 weeks of age F4,462 = 7.05 F4,231= 2.61 F14,462=0.61 WxT P=O.O8 P=0.86 development at which the exposure occurs? And does this apply equally to different window temperatures? To answer this question, we can compare hatchlings from eggs exposed to a thermal window at different times through the incubation period (i.e. excluding data on eggs that were kept in the cool incubator all the time). We analysed these data using a threefactor multivariate analysis of variance, with the factors being: (i) the population from which eggs were obtained; (ii) the thermal regime to which the eggs were exposed during their 2week window (i.e. warm or hot ); and (iii) the time at which this exposure occurred (i.e. number of weeks since commencement of incubation). These tests confirm the differences attributable to site (source population) and window temperature, as revealed by the previous analyses (see above). However, they also show that the time at which the eggs experienced higher temperatures exerted an influence on incubation periods and body sizes. The effects on morphology were evident at hatching, and also at 1 week and 2 weeks of age (Table 2). Direct effects of time of exposure were not statistically significant for locomotor speeds, but these traits were influenced by significant interactions between time of exposure and other effects (i.e. site or window temperature; see Table 2). Thus, we conclude that the timing of exposure, as well as the temperatures experienced during that period, can modify a range of phenotypic traits in hatchling lizards. ARE THERE GENERAL PATTERNS IN THE EFFECTS OF TIMING OF THERMAL EXPOSURE? Although thermal windows exerted different effects depending on the time at which they were applied, interactions between factors (Table 2) complicate interpretation. The most straightforward effect involved incubation period. Developmental rates were more sensitive to higher temperatures soon after the eggs were laid, than later in development (Fig. 3). However, this effect was evident only in eggs exposed to the hot thermal window ; incubation periods were affected much less by exposure to the warm treatment (Fig. 3). Offspring body proportions (tail length relative to SVL, and mass relative to SVL) also displayed consistent effects. For example, hatchlings from eggs that had been exposed to a hot window late in embryogenesis were heavierbodied, and had tails that were shorter relative to SVL, than their siblings that had the same thermal window earlier in development (Fig. 4). For locomotor speeds, the effects of timing of
~ 1 562 R. SHINE and M. J. ELPHICK 120. L I 2 3 4 6 7 8 Week of exposure Figure 3. Effects of a shortterm (2week) exposure to higher temperatures on the total incubation periods of scincid lizard (Bassiana duperreyi) eggs in laboratory incubators. Incubation periods were affected by the temperature during that thermal window ( 0) warm = 22 i 5 C; (0) hot = 27 f 5 Cand (especially in the hot treatment eggs) by the time in development at which the eggs were exposed to this thermal treatment. Time of exposure is shown as the week into incubation (1 to 8) at which the window was applied. Graphs show mean value i 1SE. exposure interacted strongly with window temperature (Table 2): earlier exposure to the hot window produced fasterrunning hatchlings, but the effect was less marked (and perhaps reversed) in warm window animals (Fig. 5). Looking at these results overall, there is a clear pattern for thermal exposure early (rather than late) in development to have a greater effect on hatchling phenotypes, and to generate better hatchlings. The former pattern (magnitude of effects) can be seen by comparing the responses of earlyexposed vs. lateexposed eggs to the mean values for each trait seen in eggs that were not exposed to a thermal window at all. Early exposure had more effect on hatching success, incubation period and hatchling traits than did late exposure. That is, mean values for no window eggs more closely resembled the conditions exhibited by late exposure rather than early exposure eggs (see below). A judgement as to whether earlier thermal exposure produced better hatchlings is a more subjective one, but again is supported by the shared attributes of lateexposed and nonexposed eggs. Compared to earlyexposed eggs, these groups displayed lower hatching success, a longer incubation period, shorter tails relative to SVL, and poor locomotor speed (Figs 35). Although we cannot demonstrate that these characteristics translate through into reduced organismal fitness, the implication is strong. U 2 3 2241 2oL 1 2 3 4 5 6 3 1_ 1 2 3 4 5 6 7 8 0.02 2 P LL 1 2 3 4 5 6 7 8 Week of exposure Figure 4. Effects of a shortterm (2week) exposure to higher temperatures during the incubation period on the morphology of scincid lizards (Bassiana duperreyi) hatching in laboratory incubators. Three morphological traits are shown: (A) snoutvent length (SVL), (B) tail length relative to SVL, and (C) mass relative to SVL. The latter two variables are calculated as residual scores from linear regressions. Morphological traits were affected by temperatures during the thermal window ( 0) warm = 22 5 C; (El) hot = 27 5 Cand by the time in development at which the eggs were exposed to this thermal treatment. Time of exposure is shown as the week into incubation (1 to 8) at which the window was applied. Graphs show mean value k 1 SE. DISCUSSION In highelevation environments such as those occupied by Bassiana duperreyi, weather conditions can change dramatically over very short time periods. For example,
~ LIZARD NEST TEMPERATURES 563 2 h 0.45 v E rl g 2 0.4 3 0.35 0.5 /A 0.3 3 A 0.6 E, 43 a, z 0.5 g 0.4 tn 0.3 IB 1 2 1 3 4 T 5 6 7 8 T T! 2 3 4 5 6 7 8 Week of exposure Figure 5. Effects of a shortterm (2week) exposure to higher temperatures during the incubation period on the running speeds of 1weekold scincid lizards (Bassiana duperreyi) that hatched in laboratory incubators. Running speeds were measured over two distances: (A) lm and (B) 25cm (see text for details). Locomotor performance was affected by an interaction between two factors: the temperature during that thermal window (n) warm = 22+5 C; (El) hot =27f5 Cand the time in development at which the eggs were exposed to this thermal treatment. Time of exposure is shown as the week into incubation (1 to 8) at which the window was applied. Graphs show mean value Ifr 1 SE. snow can fall even in midsummer (e.g. Green & Osborne, 1994). Our data show substantial transient variations in the thermal regimes inside lizard nests, presumably due to these shifts in weather. Remarkably, our laboratory studies show that relatively minor thermal variations (of the same magnitude as those detected in the field) can substantially modify not only the hatching success of eggs, but also the emergence dates and phenotypic traits of hatchlings. Thus, superficially trivial variation in weather patterns may have a considerable impact on the survival rates of offspring. Although we do not have field data on the ways in which hatchling phenotypes affect reproductive success in Bassiana duperreyi, the effects that we documented are likely to influence recruitment to the next i generation. For example, our brief (2week) exposure to slightly higher temperatures (the warm treatment) halved rates of mortality in the egg stage, and shortened incubation by 16%. Early hatching may substantially enhance offspring survival in such a coolclimate area (Olsson & Shine, 1997; Marco & Perez Mellado, 1998). Additionally, the thermal exposure also significantly modified a series of phenotypic traits (hatchling size, shape and speed) that are likely to affect hatchling survival (e.g. Olsson, 1992; Sinervo et al., 1992; Janzen, 1993). Thus, we doubt that these effects are biologically trivial. Presumably, Bassiana are not unique in displaying such effects. Experimental studies suggest that phenotypically plastic responses to incubation conditions are very widespread among reptiles (e.g. Rhen & Lang, 1995; Resetarits, 1996). The crucial factor, however, will be the sensitivity of embryonic reaction norms relative to the degree of thermal fluctuation in natural nests. Nests of some species of reptiles may show very little fluctuation in response to ambient weather conditions, such that embryogenesis is effectively buffered from the kinds of effects that we have documented in Bassiana. This may be especially true for species that live in the tropics (where soil temperatures may be relatively constant yearround) and species that dig very deep nests. Nonetheless, available data suggest that hatchling phenotypes depend upon nest temperatures even in species that display these traits. For example, minor thermal differences between alternative types of nestsite in one species of tropical python generate significant differences in the phenotypic traits of hatchlings (Shine et al., 1996). Why does a brief window of higherthanusual temperatures have such a marked effect on embryogenesis in Bassiana? We know too little about developmental processes in reptiles to answer this question at present, but our data identify attributes of this effect that may ultimately clarify the causal basis involved. In particular, the thermal sensitivity of all three major traits that we quantified (hatching success, developmental rate and phenotypic plasticity) showed a similar pattern of timedependence. Effects of the thermal window were higher early in embryogenesis (i.e. soon after oviposition) than they were later on (i.e. shortly before hatching). This pattern fits well with the general hypothesis that factors operating early in development can exert greater effects than those occurring later, because development proceeds by stepwise differentiation (i.e. one structure differentiates and then induces differentiation of another structure, and so forth: Gould, 1977). Similar patterns of higher sensitivity early in embryogenesis have been reported in other kinds of animals (e.g. Henry & Ulijaszek, 1996). Nonetheless, some authors have reported the reverse pattern (e.g. Congdon, Fischer & Gatten, 1995)
564 R. SHINE and M. J. ELPHICK whereas others have found no temporal shift in thermal sensitivity (e.g. Christian, Tracy & Porter, 1986). Detailed studies on fishes reveal a complex temporal sensitivity of embryogenesis (Taning, 1952). Despite this general timedependency, however, it is notable that exposure to higher temperatures continues to influence embryogenesis throughout the entire developmental period (e.g. Figs 35). This result stands in contrast to experimental results on temperaturedependent sex determination (TSD) in reptiles: the period during which incubation temperature determines offspring sex is typically brief and welldefined (e.g. Ewert & Nelson, 1991; Viets et al., 1993). This period (early to middevelopment) broadly corresponds, however, to the period of greatest thermal sensitivity of Bassiana traits (Figs 35). Because the traits we studied are continuous rather than dichotomous variables (unlike sex), we might expect to see a less clearcut pattern to their sensitivity to thermal cues. One implication of the prolonged thermal sensitivity of Bassiana embryogenesis is that occasional hot days could influence developmental trajectories, regardless of the time during incubation when such weather conditions occurred. The trend for development in Bassiana to be most sensitive to temperature soon after oviposition, also bears upon theoretical speculations on reproductive modes in squamate reptiles. One enduring puzzle in this field is the general conservatism in the duration of retention of embryos prior to oviposition: most oviparous snakes and lizards retain eggs in utem until the embryos reach stage 30 or thereabouts on the Dufaure & Hubert (1961) scale (Shine, 1983a; Blackburn, 1995). Why do we not see more interspecific diversity in this respect, with some taxa depositing eggs with very early embryos and others laying eggs with very welldeveloped embryos? Costs of prolonged egg retention to the reproducing female offer a plausible answer to the second part of this question, but the first part remains an enigma (Blackburn, 1995). A possible answer involves a greater sensitivity of embryogenesis to external factors (e.g. temperature, moisture) early in development (Shine, 1983a). If this is the case, females may be selected to retain their eggs in utem (where they can control the hydric and thermal environment) until development has passed through this sensitive phase. Apart from studies on sex determination, our data on Bassiana (e.g. Figs 35) are the first to show ontogenetic shifts in thermal sensitivity during reptilian embryogenesis. The hot treatment provides thermal conditions similar to those experienced in utem (e.g. Shine, 1980), and thus allows us to predict the effects of prolonged uterine retention on embryogenesis. Progressive increases in the duration of retention probably would have less and less effect on hatching success, incubation period and hatchling phenotypes (Figs 35). Given the increasing mass of eggs over this period (and thus, the increasing degree of physical burdening of the female), a progressive decrease in the phenotypic benefits of retention may impose strong selection on the optimal duration of uterine retention of eggs. Our study reinforces an emerging paradigm in reptilian biology: the notion that phenotypes of these animals are remarkably sensitive to physical conditions during embryogenesis. Previous work in this field has focused on variation among nests, and on the ways in which a mother s selection of nestsites can have longterm consequences for the phenotypes of her offspring. Our study extends this body of work by showing that the same kind of phenomenon can occur within a single nest over time. If our results prove to be general, then shortterm weather fluctuations may have a more important impact on reptilian populations than has generally been assumed. The phenotypic variance within a cohort of reptilesthat is, the raw material for natural selectionmay reflect not only diversity in genetic factors and in the thermal characteristics of alternative nestsites, but also the history of weather conditions over the prior incubation period. This result is both encouraging and discouraging. If indeed superficially trivial factors such as local weather conditions can exert longlasting effects on traits such as yearclass strength (and hence, age composition of the population), fieldworkers must resign themselves to the prospect that they cannot afford to ignore such processes. Environmental temperatures may thus contribute directly to phenotypic variation among hatchlings at many levels: not only among sites, and among nests within sites, but also among years, and even between successive clutches laid by the same female in the same nest in the same year. Because the sensitivity of embryogenesis to thermal regimes changes during the incubation period, nests with identical overall mean temperatures and thermal variances may nonetheless generate significantly different hatchling phenotypes. On a more encouraging note, such processes are wellsuited to experimentation in the field and the laboratory. Thus, reptiles may offer ideal model systems with which to investigate the ways in which local environments directly modify patterns of phenotypic variation in natural populations. ACKNOWLEDGEMENTS We thank S. Smith for assistance with the husbandry of eggs and hatchlings, P. Harlow and K. Vickers for technical assistance in many ways, and T. Flatt for discussions on reptilian development. This study was funded by the Australian Research Council.
REFERENCES Avery RA. 1982. Field studies of body temperatures and thermoregulation. In: Gans C, Pough FH, eds. Biology of the Reptilia, Volume 12. New York Academic Press, 93166. Avery RA, Bedford JD, Newcombe CP. 1982. The role of thermoregulation in lizard biology: predatory efficiency in a temperate diurnal basker. Behavioral Ecology and Sociobiology 11: 261267. Blackburn DG. 1995. Saltationist and punctuated equilibrium models for the evolution of viviparity and placentation. Journal of Theoretical Biology 174 199216. Bull JJ. 1980. Sex determination in reptiles. Quarterly Review of Biology 55: 421. Burger J. 1989. Incubation temperature has longterm effects on behavior of young pine snakes (Pituophis melanoleucus). Behavioral Ecology and Sociobiology 24: 201207. Burger J, Zappalorti RT. 1988. Effects of incubation temperature on sex ratios in pine snakes: differential vulnerability of males and females. American Naturalist 132 492505. Burger J, Zappalorti El!, Gochfeld M. 1987. Developmental effects of incubation temperature on hatchling pine snakes Pituophis melanoleucus. Comparative Biochemistry and Physiology 87A. 727732. Christian KA, Tracy CR, Porter WP. 1986. The effect of cold exposure during the incubation of Sceloporus undulatus eggs. Copeia 1986: 10121014. Cogger HG. 1992. Reptiles and amphibians of Australia, 4th edition. Sydney: Reed Books. Congdon JD, Fischer RU, Gatten REJ. 1995. Effects of incubation temperatures on characteristics of hatchling American alligators. Herpetologica 51: 497504. Dufaure JP, Hubert J. 1961. Table de developement du lezard vivipare: Lacerta (Zootoca) vivipara Jaquin. Archives Anatomie Micmscopie et Morphologie Experimentale 50 309328. Ewert MA, Nelson CE. 1991. Sex determination in turtles: diverse patterns and some possible adaptive values. Copeia 1991: 5G9. Gould SJ. 1977. Ontogeny and phylogeny. Cambridge, Massachusetts: Harvard University Press. Green K, Osborne W. 1994. Wildlife of the Austmlian snow country. Sydney: Reed Books. Harlow PS. 1996. A harmless technique for sexing hatchling lizards. Herpetological Review 27: 7172. Henry CJK, Ulijaszek SJ. 1996. Longterm consequences of early environment. Cambridge: Cambridge University Press. Huey R, Slatkin M. 1976. Costs and benefits of lizard thermoregulation. Quarterly Review of Biology 51: 363384. Janzen FJ. 1993. An experimental analysis of natural selection on body size of hatchling turtles. Ecology 74332 341. Marco A, PerezMellado V. 1998. Influence of clutch date LIZARD NEST TEMPERATURES 565 on egg and hatchling sizes in the annual clutch of Lacerta schreiberi (Sauria, Lacertidae). Copeia 1998 145150. Olsson M. 1992. Sexual selection and reproductive strategies in the sand lizard (Lacerta agilis). Unpublished D.Phi1. Thesis, University of Goteborg, Sweden. Olsson M, Shine R 1997. The seasonal timing of oviposition in sand lizards (Lacerta agilis): why earlier clutches are better. Journal of Evolutionary Biology 10369381. Packard GC, Packard MJ. 1988. The physiological ecology of reptilian eggs and embryos. In: Gans C, Huey RB, eds. Biology of the Reptilia, Volume 16. New York: Alan R. Liss, 523605. Packard GC, Tracy CR, Roth JJ. 1977. The physiological ecology of reptilian eggs and embryos, and the evolution of viviparity within the class Reptilia. Biological Review 52: 71105. Pengilley R 1972. Systematic relationships and ecology of some lygosomine lizards from southeastern Australia. Unpublished D.Phi1. Thesis, Australian National University, Canberra, Australia. Peterson CR, Gibson AR, Dorcas ME. 1993. Snake thermal ecology: the causes and consequences of bodytemperature variation. In: Seigel RA, Collins JT, eds. Snakes: ecology and behavior. New York McGrawHill, 241314. Resetarits WJJ. 1996. Oviposition site choice and life history evolution. American Zoologist 36: 205215. men T, Lang JW. 1995. Phenotypic plasticity for growth in the common snapping turtle: effects of incubation temperature, clutch, and their interaction. American Naturalist 146: 72G747. Shine R 1980. Costs of reproduction in reptiles. Oecologia 46: 92100. Shine R 1983a. Reptilian reproductive modes: the oviparityviviparity continuum. Herpetologica 39 18. Shine R 1983b. Reptilian viviparity in cold climates: testing the assumptions of an evolutionary hypothesis. Oecologia 57: 397405. Shine R, Elphick MJ, Harlow PS. 1997. The influence of natural incubation environments on the phenotypic traits of hatchling lizards. Ecology 7825592568. Shine R, Harlow PS. 1996. Maternal manipulation of offspring phenotypes via nestsite selection in an oviparous reptile. Ecology 77: 18081817. Shine R, Madsen T, Elphick M, Harlow P. 1996. The influence of nest temperatures and maternal thermogenesis on hatchling phenotypes of water pythons. Ecology 78 17131721. Sinervo B, Doughty P, Huey RB, Zamudio K. 1992. Allometric engineering: a causal analysis of natural selection on offspring size. Science 285 19271930. Tgning AV. 1952. Experimental study of meristic characters in fishes. Biological Review 27: 169193. Viets BE, Tousignant A, Ewer MA, Nelson CE, Crews D. 1993. Temperaturedependent sex determination in the leopard gecko, Eublepharis macularius. Journal of Experimental Zoology 265 679683.