Oecologia (1999) 119:1±8

Ó Springer-Verlag 1999

Richard Shine á Sharon J. Downes

Can pregnant lizards adjust their offspring phenotypes to environmental conditions?

Received: 6 July 1998 / Accepted: 5 December 1998

Abstract We exposed females of a highly placentotrophic viviparous scincid lizard (Pseudemoia pagenstecheri) to various environmental factors during pregnancy, and quanti®ed the e€ects of these treatments on their o€spring. The clear result was that the phenotypes of neonatal lizards can be substantially modi®ed by the environment that their mother experiences during gestation. Restricting prey availability to the females reduced the size of their o€spring. Limiting the females' basking opportunities delayed their seasonal timing of parturition, and modi®ed body proportions (tail length relative to snout-vent length) of the neonates. More surprisingly, female lizards that were regularly exposed to the scent of sympatric lizard-eating snakes gave birth to o€spring that were heavier, had unusually long tails relative to body length, and were highly sensitive to the odour of those snakes (as measured by tongue-¯ick responses). The neonates' antipredator responses were also modi®ed by the experimental treatment to which their mother was exposed. The modi®cations in body mass, tail length and response to snake scent plausibly reduce the o€spring's vulnerability to predatory snakes, and hence may constitute adaptive maternal manipulations of the neonatal phenotype. Key words Environmental maternal e€ects á Prey availability á Thermal biology á Predation á Reptile

Introduction Can reproducing animals somehow assess current environmental conditions, and use this information to modify the traits of their o€spring? Simple mathematical

R. Shine (&) á S.J. Downes Biological Sciences A08, University of Sydney, NSW 2006, Australia e-mail: [email protected] Tel: +61-2-9351-3772, Fax: +61-2-9351-5609

models (and intuition) suggest that this kind of manipulation should occur, because the optimal phenotype for an o€spring may shift from year to year in many environments (e.g. McGinley et al. 1987). Under such circumstances, selection may favor a mother's ability to manipulate the kind of o€spring she produces. Although there is good evidence that females can manipulate both the genotypes and the phenotypes of their o€spring (e.g. Clutton-Brock and Iason 1986; Charnov 1982; Olsson et al. 1996; Shine and Harlow 1996), the degree to which such manipulations are adaptive remains unclear for most cases. Presumably, some modi®cations of the o€spring phenotype (e.g. smaller babies from nutritionally stressed mothers) are simple consequences of maternal experiences rather than adaptations to enhance o€spring ®tness (Williams 1966). In other cases, there is good evidence that the response is adaptive (e.g. Mousseau 1991; Mousseau and Dingle 1991). Squamate reptiles are ideal subjects for examining these issues, because the general lack of post-hatching parental care (Shine 1988) removes complications associated with extended parental input (e.g. feeding, ``teaching''). Indeed, several previous studies on lizards and snakes have examined the ways in which maternal behaviours can in¯uence the phenotypes of o€spring (e.g. Bull et al. 1988; Burger et al. 1987; Burger 1989, 1990, 1991, 1998; Van Damme et al. 1992; Sorci and Clobert 1997). Most of this work uses oviparous species, and it is now well established that maternal nest-site selection can profoundly a€ect many aspects of the phenotype of the o€spring. The traits a€ected include body size and shape, activity levels, locomotor performance, antipredator tactics and gender (e.g. Burger et al. 1987; Burger 1989, 1990, 1991, 1998; Tousignant and Crews 1995; Shine et al. 1996). Many of the same e€ects are evident in viviparous (live-bearing) species: experimental changes to the basking opportunities for gravid lizards can a€ect the phenotypes of their o€spring (Shine and Harlow 1993; Qualls and Shine 1996; Sorci and Clobert 1997) as well as the time of year that the young are born (Schwarzkopf and Shine 1991).

2

These studies have focussed on viviparous lizard species with relatively simple placentae, and little nutrient transfer from the mother to the embryo post-ovulation (Thompson 1981). Although most live-bearing lizards display this ``ovoviviparous'' reproductive mode, substantial placental transfer of nutrients occurs during pregnancy in some squamate lineages (Stewart and Blackburn 1988). For example, recent studies have documented placentotrophy in Australian lizards of the genus Pseudemoia: approximately half of the nutrients allocated to the o€spring are transferred through the placenta during embryogenesis (Stewart and Thompson 1993). These taxa may thus o€er excellent model systems in which to investigate the ways in which reproducing females manipulate the phenotypes of their o€spring in response to environmental cues. Because they allocate nutrients progressively rather than at the beginning of embryogenesis, the opportunity exists for females to modify their allocation strategies in response to conditions that they encounter during gestation. Also, the intimate maternal-foetal connection may facilitate maternal opportunities to modify o€spring phenotypes in more subtle ways. We thus selected the highly placentotrophic species Pseudemoia pagenstecheri as our study animal, and exposed females that had recently ovulated to factors that vary in their natural habitat, in ways that might plausibly favor di€erent o€spring phenotypes. In the 1st year of the study (summer of 1996±1997), we used only two such treatments, both in a laboratory environment: controls versus food-deprived females. In the 2nd year of the study (summer of 1997±1998), we expanded the research to incorporate four treatments, in semi-natural ®eld enclosures: (1) restricted food supply (simulating year-to-year variation in the availability of insect prey); (2) restricted basking opportunities (simulating variation in weather conditions or basking opportunities); (3) frequent exposure to the scent of a predatory snake (simulating changes in the abundance of local snakes), and (4) control (ad libitum food and basking opportunities, and no snake scent). Our study was designed to assess whether or not the phenotypic traits of o€spring were modi®ed by those biologically realistic situations.

Materials and methods Study species and area P. pagenstecheri is a small (to 70 mm snout-vent length, SVL) heliothermic scincid lizard from southeastern Australia (Hutchinson and Donnellan 1992). It belongs to the entrecasteauxii species group, a morphologically conservative lineage of viviparous scincid taxa (Hutchinson and Donnellan 1992). In the population we studied at Nowendoc, New South Wales (31°35¢S, 151°45¢E), the lizards are abundant on fallen logs among tussock grassland, where they feed on small insects and other invertebrates (Hudson 1997). Their major predators are likely to be birds and elapid snakes (Drysdalia coronoides, Austrelaps superbus) (Shine 1977, 1981; R. Shine, personal observations). Skeletochronological and mark-recapture studies on Victorian populations of this taxon indicate that

females mature around 2 years of age, produce an annual litter of one to ten o€spring, and often live to an age of 5 years (Hudson 1997). Ovulation occurs in spring (September±October), and birth in summer (December±January). Because we do not know if all populations of P. pagenstecheri are placentotrophic, we collected lizards from the same population as that in which Thompson et al. (in press) documented extreme placentotrophy. Collection and maintenance of lizards We collected gravid female lizards during the latter half of October, approximately 1 month after ovulation (based on dissection of females collected over the preceding 2 months: M. B. Thompson, personal communication). The lizards were caught using baited lizard-sticks (Strong et al. 1993) and transported to Sydney, where they were measured and weighed. We then sorted the animals into the treatment groups (see Introduction) such that all relevant traits (e.g. mean body sizes, proportions of animals with regrown tails) were equivalent among the groups. In both years, this sorting was successful: there were no signi®cant di€erences among treatment groups in any maternal traits at the beginning of the studies (onefactor ANOVA, all P > 0.70). In the 1st year of the study, we housed lizards of both treatment groups (control versus food-deprived) individually in small (22 ´ 13 ´ 7 cm) plastic boxes in a room maintained at 18 ‹ 1°C. The boxes sat on heating cables, such that each female could control her body temperature over the range 24±34°C from 0900 to 1600 hours (daylight period 0700±1800 hours); outside that time the heating cables were switched o€, so that cage temperature rapidly fell to air temperature. Control females received six medium-sized crickets three times per week, whereas the ``food-deprived'' animals received half this number. To examine the possibility that a reduction in food supply might also in¯uence thermoregulation, we scored the lizards' position within their cage (distance from the heated section) three times per lizard, over a 2week period. In the 2nd year of the study (28±30 October 1997), we collected another sample of recently-ovulated female lizards from the same ®eld site. On 4 November, these lizards were placed in outdoor enclosures (ten lizards per enclosure, two enclosures per treatment) measuring 1.8 by 1.8 m, with metal walls 1.6 m high. Barkchips and concrete bricks were available for shelter, and food (3 g of live crickets) was provided every 2nd day in three of the four treatments. In the ``food'' treatment, the availability of prey was halved (i.e. 1.5 g of crickets every second day). In the ``thermal'' treatment, the time available for basking each day was reduced by erecting black shade-cloth over 85% of the appropriate pens. In the ``predator scent'' treatment, we dragged live snakes through the grass and barkchips on the ¯oor of two enclosures twice weekly, and added sandstone rocks recently used as shelter-sites by the same snakes. For this purpose, we used small elapids (white-lipped snakes, Drysdalia coronoides) that feed almost exclusively on scincid lizards (Shine 1981); the two adult female snakes used for this treatment were caught at Nowendoc when we collected the gravid lizards. Lizards from all treatment groups were brought back into the laboratory on 12 December 1997 (when it seemed that birth might be imminent), where they were maintained separately in plastic boxes as described above for the 1st year of the study. We attempted to maintain the relevant treatments under laboratory conditions by restricting basking opportunities (6 versus 12 h of heating per day) or food (3 versus 6 crickets per lizard every 2nd day), or by introducing paper scented with white-lipped snake (transferred from a snake's cage every 7th day). However, two of these treatments (predator scent and reduced feeding) ceased on 29 December, when the ®rst births occurred. Basking times continued to be restricted for the ``thermal'' group. The last births occurred on 24 January. Lizard boxes were checked at least once per day (and usually, several times per day) for the presence of neonates. The o€spring were removed immediately, weighed and measured (as was the mother), and then all of the progeny were transferred to a box set

3 up in the same way as for adults. In both years of the study, we tested locomotor speed of the young lizards at 7 days of age, using a 1-m-long by 4-cm-wide racetrack. The racetrack was kept in a constant-temperature room held at 25 ‹ 1°C. Lizards were allowed at least 60 min to reach this temperature before being tested. Each neonate was then introduced to the beginning of the trackway, and encouraged to run by gentle prodding with an artist's paintbrush. Speed was determined as the lizard crossed infrared beams (positioned at 25-cm intervals) connected to an electronic stopwatch. Thus, we could measure the animal's speed over 1 m and over the fastest 25-cm interval. In the 1st year of the study, we measured running speeds of the mothers as well as their o€spring. The mothers were tested three times: when we ®rst captured them, 81 days later (close to parturition), and 7 days after they gave birth. In the 2nd year of the study, we also tested chemoreceptive responses of the mothers and their o€spring 8 days after parturition. We presented lizards with chemical stimuli from predator and control substances on the cotton swabs of 30-cm wooden applicators (Burghardt 1967). Three conditions were presented: 1. Distilled water was used as a neutral control. Cotton tips were dipped into distilled water, and the excess blotted with tissue paper. 2. A commercial cologne (Confess) was used as a chemical pungency control. Tips were dipped into a 1:1 solution of cologne and distilled water, and excess ¯uid was blotted dry. 3. Integumentary chemicals were obtained from adult whitelipped snakes (D. coronoides). Cotton tips were dipped into distilled water and excess ¯uid was blotted dry. Tips were then rolled across the dorsal, lateral, and ventral surfaces of the snake, between the neck and abdomen (Dial et al. 1989). Lizards were tested in small plastic boxes (hatchlings, 14 ´ 10 ´ 4 cm; mothers, 22 ´ 13 ´ 7 cm) ®lled with soil to 5 mm depth. The lids were slowly removed from the boxes that contained mothers several minutes prior to the trial; the boxes that the hatchlings were tested within did not have lids. To begin a trial, one of us (SD) slowly placed the swab 1 cm anterior to the lizard's snout, and recorded the numbers of tongue-¯icks for a 60-s period beginning with the ®rst tongue-¯ick. In some cases, the lizard raised its tail and moved it rapidly from side to side (tail-raise). Our scoring system was based on the assumption that ¯eeing from the swab during the trial was a stronger response than any number of tongue-¯icks. Thus, we scored each trial as the total number of tongue ¯icks in 60 s if the lizard did not run away from the swab, and as a base unit (the greatest number of tongue-¯icks emitted by any lizard in response to any stimulus) plus (60 minus the latency to run in seconds) if the reptile ¯ed from the swab (after Burghardt 1967). The skinks were tested in a pre-set, ®xed order whereby each treatment was presented ®rst to approximately one-third of the subjects (randomly chosen). The next treatment followed the order: control, cologne, snake. Each individual was presented once with each of the three scent stimuli. All tests were performed at 25 ‹ 1°C, and at least 1 h elapsed between successive trials on the same individual. Analysis The data were checked for all relevant assumptions before statistical analysis; some variables were log-transformed to normalise variances. Initial nested ANOVA (with enclosure number nested within treatment) revealed no signi®cant di€erences among enclosures for any of the dependent variables, so we did not include ``enclosure number'' as a factor in subsequent tests. However, we incorporated clutch e€ects into our analyses because we had more than one o€spring in most litters. To do this, we used two-factor nested ANOVAs with the factors being treatment, and litter identi®cation number nested within treatment. Thus, the error term against which we tested for treatment e€ects was the ``litter within treatment'' term. This technique avoids spuriously ``signi®cant'' di€erences among treatments that are actually due to similarities

among litter-mates (Snedecor and Cochran 1987). For repeated measures of similar variables on the same animal (i.e. responses to the three scent stimuli; and running speeds over 25 cm and 1 m), we used repeated-measures analysis (incorporating the nested design as explained above). For all ANOVAs, we evaluated di€erences among treatment groups by comparing animals in the control treatment to those in each of the experimental treatments separately. For correlation analyses we used the non-parametric Spearman Rank method.

Results First summer (1996±1997) In the 1st year, we examined the e€ects of food intake. Unfortunately, many of the females aborted their o€spring, or consumed them immediately after giving birth. This was equally true of the lizards from both treatments, and post partum cannibalism continued even after we provided ad libitum food to all females. In consequence, ®nal sample sizes were small, both in terms of numbers of litters and numbers of surviving o€spring per litter. When weighed immediately after parturition, the six high-food females that produced viable o€spring had gained an average of 15.3% of their initial body mass (SD = 7.7), whereas the seven low-food females had lost an average of 11.7% (SD = 6.0; one-factor ANOVA, F1,11 = 50.68, P < 0.0001). Despite these small sample sizes, o€spring size differed strongly between the two levels of maternal nutrition. Litters of the high-food females averaged 0.20 g per o€spring (SD = 0.04), versus 0.12 g (SD = 0.04) for hatchlings from the low-food lizards (using the nested two-factor design, F1,17 = 8.84, P < 0.015). Locomotor performance relative to body size also differed between the groups. Overall mean running speeds of neonates were similar (means for high versus low = 0.54 versus 0.53 m s)1 over 1 m, 0.76 versus 0.81 m s)1 over 25 cm, repeated-measures ANOVA F1,10 = 0.18, P = 0.68), but this similarity masked two con¯icting trends. Larger o€spring tended to run faster, but the neonates from the low-food females tended to run faster relative to their (smaller) body size. This result is most easily seen from a one-factor ANCOVA (with treatment as the factor, neonatal SVL as the covariate and running speed over 1 m as the dependent variable): running speeds were a€ected by body size (F1,17 = 7.50, P < 0.015) but the form of this relationship di€ered between neonates from the two treatments (interaction F1,17 = 8.91, P < 0.01). We also detected one other di€erence in the escape tactics of the young lizards. Neonates from the low-food treatment frequently stopped abruptly, spun around 180°, elevated and wriggled their tails, and then attempted to run back in the direction from which they had come. This ``raised-tailwag'' behaviour (Shine 1995) was never seen in the neonates from ``high-food'' females (means = 0 versus 2.2 times per run; F1,9 = 7.07, P < 0.03). In contrast, experimental changes to the food intake of females did not result in any changes to their own

4

The high incidence of aborted and cannibalised o€spring during the 1st year of the study implied that the pregnant females were under high stress. Thus, we modi®ed the study to provide a more ``natural'' set of conditions. In the 2nd year, abortion, postpartum cannibalism, and

stillborn young were rare in all treatment groups (Table 1). One of the strongest correlations apparent in our data set involved the date of birth versus neonatal sizes and running speeds: o€spring born earlier in the study tended to be larger in size (correlation between birth order and neonatal SVL = )0.31, P < 0.001), but ran more slowly (q = 0.16, P0.10). This result may re¯ect the similarity in relative clutch mass (RCM, total litter mass divided by maternal post-parturient mass) between the two groups (means for high and low = 0.10 versus 0.09, F1,11 = 1.00, P = 0.76). In turn, this similarity re¯ects the fact that low-food females weighed less after giving birth (means = 2.9 versus 3.2 g), as well as producing smaller o€spring. Although we manipulated only food supply, our observations on thermoregulatory behaviour suggested that this variable was also a€ected. A two-factor nested ANOVA revealed no signi®cant di€erences among lizards within each treatment (nested factor, F42,132 = 0.87, P = 0.70), but lizards from the highfood treatment averaged closer to the heated section of the cage than did lizards from the low-food treatment (F1,42 = 5.89, P < 0.02). Second summer (1997±1998)

Trait

Mean values (SE) Control

Low-food

Low-heat

Snake scent

Maternal traits Sample size SVL (mm) Body mass at capture (g) Gestation period Response to control scent Ln response to cologne Ln response to snake scent Postparturient body mass (g) Litter size Proportion stillborn Relative clutch mass

12 55.87 3.00 70.92 1.55 1.41 1.34 3.01 4.25 0.13 0.29

(0.72) (0.13) (1.63) (0.47) (0.08) (0.14) (0.16) (0.46) (0.09) (0.04)

11 55.83 3.06 71.81 0.60 1.40 1.65 2.96 4.46 0.16 0.30

(0.83) (0.14) (1.81) (0.27) (0.17) (0.14) (0.16) (0.34) (0.09) (0.02)

14 57.05 3.04 77.79 2.14 1.46 1.45 3.23 4.71 0.06 0.30

(0.74) (0.13) (1.67)* (0.70) (0.09) (0.08) (0.15) (0.47) (0.04) (0.03)

12 55.78 2.98 71.83 1.42 1.24 1.65 3.39 4.42 0.07 0.32

(0.79) (0.13) (1.80) (0.43) (0.07) (0.09) (0.22) (0.29) (0.07) (0.02)

O€spring traits Sample size Neonatal mass (g) Neonatal SVL (mm) Neonatal tail length (mm) Response to control scent Ln response to cologne Ln response to snake scent Speed over 1 m (m s)1) Speed over 25 cm (m s)1) Date of birth (days after 1 January)

51 0.21 23.15 52.35 1.09 1.00 1.20 0.28 0.39 7.71

(0.005) (0.14) (0.39) (0.20) (0.04) (0.07) (0.01) (0.02) (0.79)

49 0.20 22.09 48.69 1.23 0.99 1.11 0.27 0.38 8.90

(0.006) (0.20) (0.88) (0.25) (0.05) (0.07) (0.01) (0.02) (0.86)

66 0.21 22.98 51.47 1.22 0.86 0.93 0.29 0.41 14.77

(0.006) (0.17) (0.49)* (0.19) (0.04) (0.06) (0.01) (0.02) (0.74)*

53 0.24 23.62 54.88 1.21 1.03 1.41 0.30 0.40 9.51

(0.006)* (0.21) (0.57)* (0.20) (0.04) (0.05)* (0.02) (0.02) (0.82)

5

(mass relative to snout-vent length) did not di€er at parturition among the di€erent treatments, nor did the females' responses to the scent of snakes, or other odours (Table 1). The lack of an enhanced chemosensory response by the adult females exposed to the snake scent was particularly interesting, because the o€spring of these females exhibited markedly higher tongue-¯ick rates when presented with snake odour (Table 1). As in the 1st year of our study, the experimental treatments did not signi®cantly modify postpartum maternal body condition, or the ratio of litter mass to maternal body mass (Table 1). The o€spring born to females in the four treatment groups di€ered signi®cantly in several respects, including their dates of birth, body sizes, relative tail lengths, and responses to the scent of predatory snakes (Table 1). Most of the e€ects that we detected were more subtle than were evident in the 1st year of the study, perhaps re¯ecting the less dramatic experimental treatments. Some of the e€ects are unsurprising: for example, parturition in females with lower basking opportunities was delayed for about 6 days (Table 1; control versus thermal treatment: F1,25 = 7.67, P < 0.011). Females with restricted food intake tended to produce smaller o€spring, although not signi®cantly so (Table 1; control versus food treatment: F1,22 = 0.05, P = 0.82). Di€erences in body shape were more clear-cut, with tail length relative to snout-vent length largest in the predatorscent neonates (control versus predator treatment: F1,21 = 8.33, P < 0.006), and smallest in the thermal o€spring (control versus thermal treatment: F1,23 = 4.74, P < 0.04; see Table 1). Mean running speeds were una€ected by any of our experimental treatments, but (as in the 1st year of the study) these treatments a€ected the frequency with which neonates displayed the raised-tail-wag antipredator response during running trials. In order to exclude artifacts due to maternal e€ects, we divided litters into two groups: those in which all o€spring displayed the raised-tail-wag behaviour at least once, and those in which at least some o€spring never displayed this behaviour. Analysis showed that this behaviour was rare in the predatorscent neonates (seen in all o€spring from only 3 of 10 litters), common in low-heat neonates (13 of 14 litters), and intermediate in frequency in the other two treatment groups (7 of 11 controls; 6 of 10 low-food; v2 = 10.2, 3 df, P < 0.02). Lastly, and most surprisingly, the o€spring from mothers exposed to the scent of predatory snakes during pregnancy showed a more intense response to the scent stimuli than did o€spring from the other treatments (Table 1; repeated measures ANOVA, control versus predator treatment: for overall level of response F1,20 = 4.43, P < 0.05). The predator-scent neonates also di€ered in the relative intensity of their response to snake scent versus the other two scents (from the same ANOVA, interaction between scent type and treatment group: F2,40 = 7.84, P < 0.002). Figure 1 shows that the general pattern of response was similar in neonates

Fig. 1 A pregnant lizard's exposure to the scent of predatory snake can in¯uence the chemosensory responses of her o€spring. The ®gure shows mean values (and associated standard deviations) of tongue¯ick rates (ln-transformed) exhibited by neonatal skinks (Pseudemoia pagenstecheri) in response to three scent stimuli: control (distilled water), pungency control (perfume), and white-lipped snake (predator). The treatment groups to which lizards were allocated were: control, limited food, limited basking opportunity, and frequent exposure to snake scent. See text for explanation and statistical tests

from each group, in that the scent of white-lipped snakes elicited greater response than cologne, which in turn elicited more response than distilled water. However, the elevation of tongue-¯ick rates in response to the snake scent was much greater in the predator-scent group than in the other neonates (Fig. 1). The most extreme behavioural response to predator-scent (tail-raise) was also more common in neonates from the snake-scent treatment (seen in o€spring from four litters, versus two litters from low-heat, and one in each of the other two treatments).

Discussion Our experiment was designed to mimic the conditions experienced by a female lizard in the wild, where there can be extensive year-to-year variation in characteristics such as weather, prey availability, and predator numbers. Our results indicate that reproducing females that experience di€erent conditions during pregnancy do indeed produce o€spring with di€erent phenotypes, at least in the semi-natural enclosures used in our study. Although it is plausible that some of those modi®cations may enhance o€spring ®tness, parsimony suggests that most of the e€ects that we observed re¯ect direct (nonadaptive) responses of the females and their embryos to changed conditions. For example, the delayed parturition of females denied access to prolonged basking opportunities presumably re¯ects the dependence of developmental rates on temperature, and there is no need to attribute any adaptive signi®cance to such an

6

e€ect (Schwarzkopf and Shine 1991). Similarly, the smaller size of o€spring from food-deprived mothers is more parsimoniously explained as a direct consequence of lower maternal nutrition, rather than as evidence of adaptive modi®cation. An extensive literature on viviparous mammals demonstrates that neonatal size is reduced if maternal food intake is curtailed (e.g. Rasmussen and Fischbeck 1987). Data are more limited for reptiles, and suggest that o€spring size is a€ected by maternal feeding during gestation in some viviparous taxa (e.g. Doughty and Shine 1998; X. Bonnet, personal communication; P. Weatherhead, personal communication) but not others (e.g. Rohr 1997; Gregory and Skebo 1998). Similarly, studies on oviparous lizards indicate that reducing energy stores via tail autotomy reduces o€spring size in some species (Dial and Fitzpatrick 1981) but not others (Taylor 1984). Although these responses to lower food availability and lower basking opportunities may be direct e€ects rather than adaptations, this does not mean that they lack biological signi®cance. Several studies on lizards have demonstrated that the body size of a neonate can a€ect its probability of survival (e.g. Fox 1975; Sinervo et al. 1992), and a growing literature suggests that hatchling ®tness is strongly in¯uenced by the time of year at which the o€spring is produced (e.g. Olsson and Shine 1997; Marco and Perez-Mellado 1998). Thus, environmentally induced modi®cations to the o€spring phenotype may well have considerable consequences in microevolutionary terms, even if that ¯exibility results from direct e€ects rather than adaptive pathways. Trade-o€s between traits may constrain the ability of females to produce ``optimal'' phenotypes in their o€spring. For example, neonatal size and running speed covaried with date of birth in our results: o€spring born earlier were larger, but ran more slowly relative to body size. This pattern may re¯ect an underlying trade-o€, whereby reproducing females that retain their o€spring for longer can thereby produce neonates with higher locomotor performance, but at a smaller size because of the additional metabolic expenditure during prolonged gestation. Presumably, the advantages of early birth may outweigh those of enhanced locomotor performance under some conditions, but not others. The underlying mechanisms involved may be similar to those in oviparous skinks in which o€spring incubated at higher temperatures (and thus, likely to hatch early in the season) display superior locomotor performance (Elphick and Shine 1998). Our most interesting results involve environmentally induced modi®cations that are not easily explicable as direct e€ects. In these cases, females subjected to particular environmental conditions produced neonates with phenotypes that might confer high ®tness in that environment. Our study provides three potential examples of such phenomena. First, locomotor performance of the neonates relative to o€spring size di€ered between the two treatments in the 1st year of the study: the small o€spring (from food-deprived females) were faster run-

ners relative to their body size than were the larger o€spring from the control females. The second case involves relative tail lengths of the neonates in the 2nd year of the study. Females exposed to predator scent produced neonates with unusually long tails (relative to SVL) whereas the reverse was true for females from the low-basking treatment. In both cases, there is a plausible link to ®tness. A longer tail may enhance the e€ectiveness of tail autotomy in an encounter with a predator (Dial and Fitzpatrick 1984; Greene 1988), whereas tail length may be the ®rst component of the body to be ``sacri®ced'' if there is insucient time to build an optimally sized o€spring. The third and most interesting case involves the response of neonatal lizards to the scent of a predatory snake. Remarkably, mothers exposed to that scent through pregnancy produced o€spring that were approximately twice as responsive to the scent as were neonates from the other treatments (Fig. 1), as well as being heavier and relatively longertailed (Table 1). What mechanism might be responsible for the greater sensitivity of the ``predator-scent'' o€spring to snake odours in the present study? Presumably, this e€ect did not operate via changes to either maternal nutrition or thermoregulation. Both of these traits could plausibly be a€ected by predator scent (Van Damme et al. 1990), and hence could modify o€spring phenotypes via this route. However, neither food deprivation nor restricted basking opportunities increased the o€spring's response to snake odour in our experiments (Table 1). Nonetheless, at least three possible mechanisms are consistent with our data: 1. The neonates may have detected faint traces of snake scent in their mothers' cage soon after birth, and this exposure may have primed them to later respond vigorously to the same scent. However, we doubt this explanation because: (a) the neonates were generally removed from their mother's cage within a few hours of parturition; (b) snake scent was not added to any cages after the ®rst birth occurred on 29 December; (c) the females exposed repeatedly to snake scent did not thereby modify their own level of response; and (d) there was no trend for ``snake-scent'' o€spring born later in the year to respond less strongly to the scent (relative to neonates in other treatments). 2. The mothers may have been highly stressed by frequent exposure to predator scent during pregnancy, and this stress somehow modi®ed developmental trajectories of the o€spring such that they exhibited increased levels of responsiveness to strange odours. There is good evidence that stress during pregnancy can a€ect embryogenesis in mammals (e.g. Glavin 1984). Perhaps the same can occur with placentotrophic reptiles. 3. The mothers somehow manipulated the behavioural phenotypes of their o€spring such that the neonates were ``primed'' to respond to the scent of the same snakes that their mothers had encountered (olfactorily) during pregnancy. At ®rst sight, this scenario

7

sounds uncomfortably close to Lamarckian evolution, or the common urban myth that human babies are frightened by whatever frightened their mother during pregnancy. However, the process we documented in our lizards does not require genetic changes in the o€spring; simply a modi®cation of the behavioural phenotype. We know too little about brain function to speculate about the ways in which such modi®cations might occur. Regardless of the mechanism, our results clearly suggest that the phenotypes of neonates born to the mothers in the predator-scent treatment di€ered substantially from those produced by mothers in the control treatment (Table 1). This is an intriguing result, and worth additional study. More generally, the diversity of reproductive modes among scincid lizards o€ers an ideal opportunity for researchers to tease apart the ways in which environmental cues experienced by reproducing females may modify the phenotypes of their o€spring. Because oviparous taxa are ``capital breeders'' (Drent and Daan 1980), the primary mechanism by which they can modify their o€spring's phenotypes involves selection of nest-sites (and thus, incubation conditions: e.g. Shine and Harlow 1996). Viviparous species have a longer ``window of opportunity'' to a€ect developmental trajectories of their o€spring. In the case of fully ``ovoviviparous'' (non-placentotrophic) live-bearers, that control might be exerted mostly by maternal thermoregulatory behaviour (e.g. Beuchat and Ellner 1987; Shine and Harlow 1993). The evolution of complex placentation and matrotrophy introduces the possibility for prolonged and subtle maternal manipulations of the o€spring, because these ``income breeders'' are able to modify their investment patterns in response to conditions that they encounter during pregnancy. Our study is a preliminary one, but it supports the notion that the conditions experienced by reproducing female reptiles can generate signi®cant variation in the phenotypes of their o€spring. Acknowledgements We thank M. Elphick, A. Elphick, K. Vickers, S. Smith and P. Harlow for lizard husbandry, data acquisition, and numerous discussions. S. Nauwelaerts and C. Austin helped collect the lizards, and S. Ramsay released them at Nowendoc after the work was completed. The research was supported ®nancially by the Australian Research Council.

References Beuchat CA, Ellner S (1987) A quantitative test of life history theory: thermoregulation by a viviparous lizard. Ecol Monogr 57:45±60 Bull JJ, Gutzke WHN, Bulmer MG (1988) Nest choice in a captive lizard with temperature-dependent sex determination. J Evol Biol 2:177±184 Burger J (1989) Incubation temperature has long-term e€ects on behavior of young pine snakes (Pituophis melanoleucus). Behav Ecol Sociobiol 24:201±207 Burger J (1990) E€ects of incubation temperature on behavior of young black racers (Coluber constrictor) and kingsnakes (Lampropeltis getulus). J Herpetol 24:158±163

Burger J (1991) E€ects of incubation temperature on behavior of hatchling pine snakes: implications for reptilian distribution. Behav Ecol Sociobiol 28:297±303 Burger J (1998) Antipredator behaviour of hatchling snakes: e€ects of incubation temperature and simulated predators. Anim Behav 56:547±553 Burger J, Zappalorti RT, Gochfeld M (1987) Developmental e€ects of incubation temperature on hatchling pine snakes Pituophis melanoleucus. Comp Biochem Physiol 87A:727±732 Burghardt GM (1967) Chemical-cue preferences of inexperienced snakes: comparitive aspects. Science 157:718±721 Charnov EL (1982) The theory of sex allocation. Princeton University Press, Princeton Clutton-Brock TH, Iason GR (1986) Sex ratio variation in mammals. Q Rev Biol 61:339±374 Dial BE, Fitzpatrick LC (1981) The energetic cost of tail autotomy to reproduction in the lizard Coleonyx brevis (Sauria: Gekkonidae). Oecologia 51:310±317 Dial BE, Fitzpatrick LC (1984) Predator escape success in tailed versus tailless Scincella lateralis (Sauria: Scincidae). Anim Behav 32:301±302 Dial BE, Weldon PJ, Curtis B (1989) Chemosensory identi®cation of snake predators Phyllorhynchus decurtatus by banded geckos Coleonyx variegatus. J Herpetol 23:224±229 Doughty PD, Shine R (1998) Reproductive energy allocation and long-term energy stores in a viviparous lizard (Eulamprus tympanum). Ecology 79:1073±1083 Drent RH, Daan S (1980) The prudent parent: energetic adjustments in avian breeding. In: Klomp H, Woldendorp JW (eds) The integrated study of bird populations, North-Holland, Amsterdam, pp 225±252 Elphick MJ, Shine R (1998) Longterm e€ects of incubation temperatures on the morphology and locomotor performance of hatchling lizards (Bassiana duperreyi, Scincidae). Biol J Linn Soc 63:429±447 Fox SF (1975) Natural selection on morphological phenotypes of the lizard Uta stansburiana. Evolution 29:95±107 Glavin GB (1984) Prenatal maternal stress: di€erential e€ects upon male and female o€spring responses to restraint stress as adults. Pavlovian J Biol Sci 19:157±159 Greene HW (1988) Antipredator mechanisms in reptiles. In: Gans C, Huey RB (eds) Biology of the Reptilia, vol 16. Alan R Liss, New York, pp 1±152 Gregory PT, Skebo KM (1998) Trade-o€s between reproductive traits and the in¯uence of food intake during pregnancy in the garter snake, Thamnophis elegans. Am Nat 151:477±486 Hudson SC (1997) Patterns of life history, growth and morphology in south-eastern Australian viviparous skinks. La Trobe University, Bundoora, Victoria Hutchinson MN, Donnellan SC (1992) Taxonomy and geographic variation in the Australian lizards of the genus Pseudemoia (Scincidae: Lygosominae). J Nat Hist 26:215±264 Marco A, Perez-Mellado V (1998) In¯uence of clutch date on egg and hatchling sizes in the annual clutch of Lacerta schreiberi (Sauria, Lacertidae). Copeia 1998:145±150 McGinley MA, Temme DH, Geber MA (1987) Parental investment in o€spring in variable environments: theoretical and empirical considerations. Am Nat 130:370±398 Mousseau TA (1991) Geographic variation in maternal-age e€ects on diapause in a cricket. Evolution 45:1053±1059 Mousseau TA, Dingle H (1991) Maternal e€ects in insect life histories. Annu Rev Entomol 36:511±534 Olsson M, Shine R (1997) The seasonal timing of oviposition in sand lizards (Lacerta agilis): why earlier clutches are better. J Evol Biol 10:369±381 Olsson M, Shine R, Gullberg A, Madsen T, Tegelstrom H (1996) Female lizards control paternity of their o€spring by selective use of sperm. Nature 383:585 Qualls CP, Shine R (1996) Reconstructing ancestral reaction norms: an example using the evolution of reptilian viviparity. Funct Ecol 10:688±697

8 Rasmussen KM, Fischbeck KL (1987) E€ect of repeated reproductive cycles on pregnancy outcome in ad-libitum fed and chronically food-restricted rats. J Nutrition 117:1959±1966 Rohr DH (1997) Demographic and life-history variation in two proximate populations of a viviparous skink separated by a steep altitudinal gradient. J Anim Ecol 66:576±578 Schwarzkopf L, Shine R (1991) Thermal biology of reproduction in viviparous skinks, Eulamprus tympanum: why do gravid females bask more? Oecologia 88:562±569 Shine R (1977) Habitats, diets and sympatry in snakes: a study from Australia. Can J Zool 55:1118±1128 Shine R (1981) Venomous snakes in cold climates: ecology of the Australian genus Drysdalia (Serpentes: Elapidae). Copeia 1981:14±25 Shine R (1988) Parental care in reptiles. In: Gans C, Huey RB (eds) Biology of the Reptilia, vol 16. Alan R Liss, New York, pp 275± 330 Shine R (1995) A new hypothesis for the evaluation of viviparity in reptiles. Am Nat 145:809±823 Shine R, Harlow P (1993) Maternal thermoregulation in¯uences o€spring viability in a viviparous lizard. Oecologia 96:122±127 Shine R, Harlow PS (1996) Maternal manipulation of o€spring phenotypes via nest-site selection in an oviparous reptile. Ecology 77:1808±1817 Shine R, Madsen T, Elphick M, Harlow P (1996) The in¯uence of nest temperatures and maternal thermogenesis on hatchling phenotypes of water pythons. Ecology 78:1713±1721 Sinervo B, Doughty P, Huey RB, Zamudio K (1992) Allometric engineering: a causal analysis of natural selection on o€spring size. Science 285:1927±1930 Snedecor GW, Cochran WG (1987) Statistical methods, 7th edn. Iowa State University Press, Ames

Sorci G, Clobert J (1997) Environmental maternal e€ects on locomotor performance in the common lizard Lacerta vivipara. Evol Ecol 11:531±534 Stewart JR, Blackburn DG (1988) Reptilian placentation: structural diversity and terminology. Copeia 1988:838±851 Stewart JR, Thompson MB (1993) A novel pattern of embryonic nutrition in a viviparous reptile. J Exp Biol 174:97±108 Strong D, Leatherman B, Brattstrom BH (1993) Two new simple methods for catching small fast lizards. Herpetol Rev 24:22±23 Taylor JA (1984) Ecology of the lizard Ctenotus taeniolatus: Interaction of life history, energy storage and tail autotomy. PhD thesis, University of New England, Armidale, Australia Thompson J (1981) A study of the sources of nutrients for embryonic development in a viviparous lizard, Sphenomorphus quoyii. Comp Biochem Physiol A, Comp Physiol 70:509±518 Thompson MB, Stewart JR, Speake BK, Russell KJ, McCartney RJ, Surai PF (in press) Placental nutrition in a viviparous lizard with a complex placenta. J Zool Tousignant A, Crews D (1995) Incubation temperature and gonadal sex a€ect growth and physiology in the leopard gecko (Eublepharis macularius), a lizard with temperature-dependent sex determination. J Morphol 224:159±170 Van Damme R, Bauwens D, Vanderstighelen D, Verheyen RF (1990) Responses of the lizard Lacerta vivipara to predator chemical cues: the e€ects of temperature. Anim Behav 40:298±305 Van Damme R, Bauwens D, Brana F, Verheyen RF (1992) Incubation temperature di€erentially a€ects hatching time, egg survival and sprint speed in the lizard Podarcis muralis. Herpetologica 48:220±228 Williams GC (1966) Adaptation and natural selection: a critique of some current evolutionary thought. Princeton University Press, Princeton