Offspring performance and the adaptive benefits of. prolonged pregnancy: experimental tests in a viviparous lizard

Functional Ecology 2009, 23, 818 825 doi: 10.1111/j.1365-2435.2009.01544.x Offspring performance and the adaptive benefits of Blackwell Publishing Ltd prolonged pregnancy: experimental tests in a viviparous lizard Geoffrey M While 1 *, Tobias Uller 2 and Erik Wapstra 1 1 School of Zoology, Private Bag 05, University of Tasmania, Tasmania 7001, Australia; and 2 Edward Grey Institute, Department of Zoology, University of Oxford, Oxford OX1 3PS, UK Summary 1. Offspring locomotor performance has been shown to influence fitness related traits in a wide range of taxa. One potential mechanism by which viviparous animals can increase the performance (e.g. sprint speed) of their offspring is by prolonging pregnancy (beyond that required for complete development). However, to date studies examining this potentially important maternal effect have been largely descriptive. 2. The skink Egernia whitii is an ideal candidate species to examine the consequences of delayed parturition on the performance of offspring as it routinely gives birth asynchronously despite synchronous offspring development. 3. Using correlative data from a natural population and experimental manipulations of birthing asynchrony, we tested the prediction that, within litters, last born offspring have a better locomotor performance than first born offspring. 4. We show that prolonged pregnancy does significantly influence average offspring locomotor performance; however, contrary to predictions, the direction of this effect is dependent on gestation length and thus offspring date of birth. Last born offspring had significantly poorer performance than first born offspring in litters early in the season with this pattern reversed late in the season. 5. These results do not support the hypothesis that prolonged retention of fully formed offspring consistently increases offspring performance; however, they may help us understand the asymmetries in offspring competitive ability generated by birthing asynchrony. Key-words: Birth date, birthing asynchrony, locomotor performance, maternal effects, sibling rivalry Introduction Locomotor performance has been shown to influence fitness in a wide range of taxa (Watkins 1996; Olsson & Shine 2003; Le Galliard et al. 2004; Miles 2004; Walker et al. 2005; Husak et al. 2006). Specifically, in reptiles, offspring performance at birth or hatching has been shown to be an important predictor of growth and survival in natural populations (Warner & Andrews 2002; Le Galliard et al. 2004; Miles 2004). For example, lizards that sprint faster commonly have higher survival (Warner & Andrews 2002; Miles 2004; Irschick & Meyers 2007), suggesting strong selection on the performance of juveniles. Offspring traits are also commonly affected by maternal effects (Mousseau & Fox 1998; Uller 2008). Thus, we should also expect strong selection on maternal effects *Correspondence author. E-mail: gwhile@utas.edu.au that influence offspring performance during both pre- and post-natal development. In line with this, maternal effects on offspring performance have been described in a number of oviparous and viviparous lizard species (e.g. Sorci & Clobert 1997; Swain & Jones 2000; Meylan & Clobert 2004; reviewed in Deeming 2002; but see Uller & Olsson 2003, 2006). In one example, Shine and Olsson (2003) found that in Niveoscincus microlepidotus, a viviparous skink species that facultatively retain offspring beyond that required for complete development (Olsson & Shine 1998), offspring locomotor performance was correlated with the length of time fully formed offspring were retained within the oviduct. Offspring whose parturition was delayed to a greater extent had faster sprint speeds than offspring who were retained for a shorter period of time (Shine & Olsson 2003). Plausible mechanisms for these patterns include prenatal development of musculoskeletal or brain function and the time required to complete reversion 2009 The Authors. Journal compilation 2009 British Ecological Society

Delayed parturition and locomotor performance 819 from intra- to extra-uterine environmental conditions (Shine & Olsson 2003, see also Vince & Chinn 1971; Woolf et al. 1976). Thus, Shine and Olsson (2003) suggested that prolonged pregnancy, beyond that required for complete development, in viviparous reptiles and other species could be an adaptive strategy to enhance offspring performance, and that selection would favour flexibility in this trait. Suggestive as these results are, the current status of the adaptive prolonged pregnancy hypothesis is largely based on correlative data. As prolonged retention of offspring also results in a later date of birth, it is impossible to separate the direct effects of prolonged intra-uterine (or intra-oviductal) retention of offspring from potential direct and indirect effects on birth date relative to the population mean (Shine & Olsson 2003). For example, mothers delaying parturition of fully formed young may also invest relatively more into current reproduction or choose relatively cooler basking conditions that have combined positive effects on date of parturition and offspring performance. In order to overcome these problems it is important that we identify a system in which the direct effects of facultative prolonged retention of embryos can be separated from the direct and indirect effects of birth date. Unlike other reptiles, the lizard Egernia whitii (Lacepede, 1804) normally gives birth asynchronously (with birth of offspring within a litter spread over up to 9 days) despite that offspring development is synchronous (While et al. 2007). Thus, while within litter conditions during development are equivalent for all siblings, there are between sibling differences in the extent to which parturition is prolonged. Because of this, E. whitii offers an opportunity in which to directly test the adaptive prolonged pregnancy hypothesis by using a combination of correlative and experimental approaches to examine the relative performance of first and last born offspring within a litter. Importantly, if indeed there is a difference in performance between first and later born offspring, this could form a basis for understanding the establishment of the sibling hierarchies created by asynchronous birth within this species (While & Wapstra 2008). Our study is further facilitated by a number of factors. First, there is substantial variation among females in the length of gestation in the wild, which means that the difference in birth date between first and last offspring within litters is minor compared to the large variation among litters (While et al. 2007). Second, there is also variation among females in the degree of asynchrony, which allows comparisons between the first and later born offspring within females directly in relation to the length of the birthing asynchrony. Finally, we can experimentally manipulate both of the above factors; the extent to which parturition is prolonged within a litter using hormonal induction of parturition (While et al. 2007; While & Wapstra 2008), and the length of gestation using maternal thermal regime (G.M. While, unpublished data, see also Wapstra 2000; Shine & Harlow 1993). This allows us to examine the relative effect of both delayed parturition (within litters) and offspring birth date (as a consequence of gestation length) on offspring locomotor performance. Based on Shine and Olsson (2003), Fig. 1. An adult female White s skink, Egernia whitii. we predict that offspring whose parturition has been prolonged beyond that required for development (the last born offspring) will have superior locomotor performance compared to offspring whose parturition has not been delayed (first born offspring), and that these patterns will be independent of birth date. Materials and methods STUDY SPECIES Egernia whitii is a medium sized (up to 100 mm snout-vent length) viviparous lizard found throughout a broad range in southeastern Australia (Fig. 1). We used skinks from a population on the East Coast of Tasmania, Australia (42 57 S, 147 88 E). Males and females are sexually monomorphic, become reproductively mature at approximately 3 years, and display an overall life span of 9 10 years (Chapple 2003; While et al. 2009). Reproduction occurs annually, with mating occurring during the Austral spring (September October), and females giving birth to offspring in the Austral summer (January February) following a 3 4-month gestation (Chapple 2003; While et al. 2007, 2009). Egernia whitii have a relatively simple placenta allowing for limited maternal provisioning throughout gestation (Weekes 1935). Average birth dates for females in this population are variable across years, but occur during the Austral summer (January March). Birthing asynchrony has previously been documented in this species, occurring in 100% of litters (While et al. 2007). In the field females give birth to offspring with an average of 2 days between births; however, spread of births varies both between litters within years and in the mean spread between years (While et al. 2007). FIELD AND LABORATORY PROTOCOL We tested the adaptive prolonged pregnancy hypothesis in three separate ways. In all cases, locomotor performance was tested immediately following birth using a 1-m long, 5-cm wide sprint speed track. Before testing, all offspring were kept in a sealed container within a water bath held at 27 ± 0 1 C for a minimum of 20 min to ensure that all individuals were tested at their preferred body temperature. Each offspring was then introduced to the beginning of the sprint speed track and encouraged to run by the tapping of its tail with a paint brush (Wapstra 2000; Warner & Shine 2006). Speed

820 G. M. While et al. was determined as lizards crossed four infrared beams (20 cm apart) connected to an electronic stopwatch. Thus, speed over both the total race track (i.e. average speed) as well as speed over any of the individual components (i.e. best) was measured. Any trials in which offspring refused to run, attacked the brush, or ran in the wrong direction were removed from the analysis (Losos et al. 2002). Natural population patterns We first examined natural patterns of offspring locomotor performance with respect to an individual s birth order (first vs. last born) and date of birth. Data collection was carried out over three reproductive seasons (2004/05 to 2006/07) in a population with individually marked animals, and followed identical protocols in each season (see also While et al. 2007, 2009). At the end of gestation, all pregnant females were brought into specifically designed terrestrial ecology facilities at the University of Tasmania to give birth. Females were housed individually in plastic terraria (30 60 40 cm) in a room maintained at an ambient temperature of 17 C throughout the day and 10 C at night. Each terrarium was supplied with a 40 W spotlight suspended 15 cm above a basking rock (providing a thermal gradient of 40 17 C in the terraria), food (Tenebrio larvae, crushed fruit) and water were available ad libitum. Basking lights were set on a timer to allow females to bask to their preferred body temperature for 10 h per day. At birth, offspring were temporarily removed from their mother to be marked and have their weight (±1 mg), snout-vent length, and total length (± 1 mm) recorded. All offspring were then sprinted (see above methods) and then released with their mother, at their mother s site of capture, within 3 days of birth. Experimental induction of birth In the second field season (2005/06), we experimentally manipulated asynchrony of birth by inducing birth of second and third born offspring at birth of the first offspring. This allowed us to compare offspring locomotor performance between offspring whose parturition was delayed (as normal) and offspring whose birth was induced. To achieve this, 82 females were brought into the laboratory in the middle of gestation (mid December), and kept under identical conditions to females from the main study population (see above). Females were then divided into two treatment groups; in the first treatment group (n = 46) females were allowed to give birth naturally and therefore asynchronously. In the second treatment group (n = 36) females gave birth to the first offspring within a litter naturally (cages were checked every hour for birth of offspring during each day). Following birth of the first offspring, females were given an intraperitoneal injection of the birth hormone arginine vasotocin (AVT) to induce birth of remaining offspring (Cree & Guillette 1991; While et al. 2007). AVT was administered at a dose of 1 μg g 1 body weight dissolved in 100 μl of lizard ringer solution, based on modified protocols from Atkins et al. (2006). After the AVT injection, females were monitored every 30 min for subsequent births; if no births resulted within 2 h of administration, a second dose of AVT was administered (32 females required a single dose of AVT, 4 required a second dose of AVT). Following birth, all offspring were temporarily removed from their mother to be marked and have their birth date, weight (±1 mg), snout-vent length, and total length (±1 mm) recorded. Offspring were then sprinted (see the above methods) before being returned to their mother s terrarium and later returned, with their mother, to their mother s site of capture. Experimental prolonging of parturition In the third field season (2006/2007), we experimentally manipulated the basking conditions a female experienced during gestation to examine the effect of thermal regime during gestation (and its subsequent effects on gestation length and within litter spread of births; G.M. While, unpublished data) on offspring locomotor performance. To achieve this, 44 females were brought into the laboratory at the beginning of gestation (October). Females were assigned into two treatment groups representing different thermal regimes, each allowing them to bask to their preferred body temperature of 30 C for a different length of time; with 22 females kept under a short basking regime (4 h access to heat lamp per day) and 22 females kept under a long basking treatment (10 h access to heat lamp per day). These conditions represent annual variation in temperature typically encountered by skinks in the natural population (Wapstra 2000; Wapstra et al. 1999). All individuals were maintained under these conditions until offspring were born, with all other laboratory conditions identical to those for housing females from the natural population (see above). At the end of gestation, terraria were checked hourly for birth of offspring. At birth, birth date, offspring weight (±1 mg), snout-vent length, and total length (±1 mm) were recorded. Offspring were then sprinted (see the above methods) before being returned to their mother s terrarium and later returned, with their mother, to their mother s site of capture. STATISTICAL ANALYSIS Average sprint speed (over the whole of the track) and best sprint speed (fastest of the four times) were highly correlated with one another (P = 0.0001, r = 0 86). As average sprint speed has previously been shown to be highly repeatable over a three month period in E. whitii (ρ = 0 81, F (23,23) = 5 29, P < 0 001; Sinn et al. 2008), we used average sprint speed in all analyses. For each analysis we first looked at whether there were differences in individual locomotor performance between first and last born offspring. We then addressed whether there were shifts in the magnitude of this difference between first and last born offspring (i.e. ignoring second born offspring in litters of three), both within and between seasons/treatments. To achieve this latter aim we created a new variable; within litter locomotor difference (i.e. difference in performance between the first and last born offspring), calculated by subtracting the sprint speed of the first born offspring from that of the last born. In all cases sprint speeds are measured in metres per second, thus high values represent a faster sprint speed. Analyses of differences in locomotor performance of the first and last born offspring were conducted using general linear mixed models (PROC MIXED). In all cases an individual s average sprint speed was entered as the dependent variable, with birth order (first or last), season (only for the natural population) and treatment (only for the experimental treatments) entered as fixed factors, litter birth date (based on the date of birth of the first offspring within a litter), offspring mass at birth, and litter size entered as covariates, and clutch identification entered as a random repeated factor. Analysis of within litter locomotor differences was conducted using general linear models (PROC GLM). In all cases the within litter locomotor difference was entered as the dependent variable, with season (only for the natural population), and treatment (only for the experimental treatments) entered as fixed factors, and litter birth date, average spread between births (only for the natural population as spread was manipulated within experimental treatments), and litter size entered as covariates. Because of significant differences between treatments

Delayed parturition and locomotor performance 821 Table 1. Litter characteristics for females within the three experimental treatments. All estimates are for females who gave birth to litters of two or three, excluded are litter sizes of one. The percentage of litters in which female gave birth to three offspring is represented by the increment in average litter size in excess of 2 (e.g. 33% of litters born within the natural population contained three offspring). Birth date represents the average date of birth of all litters born within each treatment. Average spread represents the average time, in days, between the births of subsequent offspring, thus it takes into account differences in litter size between treatments Experiment Treatment Number of litters Litter size Average offspring mass Birth date Average spread Natural Natural population 69 2 33 ± 0.05 1 34 ± 0 01 28th January 2 94 ± 0 19 (1 9) Experimentally induced Synchronous 33 2 21 ± 0.08 1 26 ± 0 02 6th February 0 Asynchronous 37 2 29 ± 0.08 1 34 ± 0 02 2nd February 2 58 ± 0 25 (1 6) Experimentally prolonged Short basking (4 h) 14 2 14 ± 0.10 1 26 ± 0 03 19th March 1 82 ± 0 36 (0 5) Long basking (10 h) 20 2 25 ± 0.09 1 37 ± 0 01 6th February 2 65 ± 0 51 (1 9) in offspring date of birth and mass at birth in the thermal experiment (G.M. While, unpublished data), litter birth date and mass at birth were not entered as covariates in analysis of sprint speeds for the thermal experiment. For the thermal experiment we also analysed the mean level differences between treatments in overall average locomotor performance using a general linear mixed model (PROC MIXED). Average sprint speed was entered as the dependent variable, treatment was entered as a fixed factor, and clutch identification was entered as a random repeated factor. All analyses were carried out using sas stat v.9 1. For all mixed models, significance of fixed effects were tested using F-tests, with the degrees of freedom calculated using the Satterthwaite s approximation (Littell et al. 1996). All models started with the full model including all relevant interaction terms and we subsequently eliminated non-significant (P > 0 05) interaction terms. We report results for models containing all main effects and significant interaction terms following backward elimination. All data were checked for violation of assumptions, including homogeneity of slopes where covariates were used. Results NATURAL POPULATION PATTERNS Over the three reproductive seasons, a total of 183 offspring were born to the 91 females that were brought into the laboratory to give birth. Of these, 26 offspring were excluded from the analysis because they refused to run, attacked the brush, or ran in the wrong direction, 22 offspring were excluded because they came from litters with single offspring, and 2 offspring were not sprinted resulting in a final sample of 134 offspring (Table 1). There were significant differences in locomotor performance between first and last born offspring, with first born offspring having a significantly faster average sprint speed than last born offspring on average (first born 0 32 ± 0 02, last born 0 26 ± 0 01; Table 2). This pattern was consistent across years despite yearly variation in the average locomotor performance (Table 2). We neither found any difference in mass between first and last born offspring (F 1,54.8 = 2 99, P = 0 09) nor found an effect of offspring mass on an individuals locomotor performance (Table 2). There were also significant yearly shifts in the within litter locomotor difference between first and last born offspring Table 2. Factors contributing to variation in offspring locomotor performance for Egernia whitii offspring within a natural population, and within experimental populations where birth was (a) experimentally induced, and (b) experimentally prolonged. [Interaction terms which did not alter the results of the main effects (i.e., P-values greater than 0 05) are not shown, values in bold face represent statistically significant effects] Experiment Predictor variables (Table 3), as well as a weak, but statistically significant, shifts with birth date (Table 3). Litters born early in the season had a positive litter locomotor difference (i.e. first born offspring within a litter had faster sprint speed than last born offspring), whereas litters born late in the season had a negative within litter locomotor difference (i.e. last born offspring within a litter had a faster sprint speed faster sprint speed then first born offspring). There was no link between the within litter locomotor difference and the extent of spread between births (Table 3). EXPERIMENTAL INDUCTION OF BIRTH Average locomotor performance (m s 1 ) Natural Season F 2,127 = 5 98 P = 0 0033 Birth Order F 1,127 = 8 12 P = 0 0051 (First or Last) Birth Date F 1,127 = 3 12 P = 0 0798 Mass at Birth F 1,127 = 0 32 P = 0 5755 Litter Size F 1,127 = 0 45 P = 0 5038 Experimentally Treatment F 1,66.4 = 4 91 P = 0 0301 induced Birth Order F 1,68.1 = 0 13 P = 0 7204 (First or Last) Birth Date F 1,74.4 = 13 34 P = 0 0005 Mass at Birth F 1,103 = 2 50 P = 0 1169 Litter Size F 1,74.4 = 0 49 P = 0 4849 Experimentally Treatment F 1,28.1 = 8 39 P = 0 0072 prolonged Birth Order F 1,24.3 = 3 57 P = 0 0709 (First or Last) Treatment F 1,24.3 = 10 00 P = 0 0042 Birth Order Litter Size F 1,29.6 = 0 01 P = 0 9597 A total of 82 females were brought into the laboratory to give birth which resulted in 156 offspring, with 70 offspring born

822 G. M. While et al. Table 3. Factors contributing to variation in within litter locomotor difference for Egernia whitii offspring within a natural population and within experimental populations where birth was (a) experimentally induced, and (b) experimentally prolonged. [Interaction terms which did not alter the results of the main effects (i.e. P-values > 0 05) are not shown, values in bold face represent statistically significant effects] Experiment Predictor Variables Litter Locomotor Difference (m s 1 ) Natural Season F 2,49 = 4 91 P = 0 0119 Birth date F 1,49 = 4 14 P = 0 0479 Birth spread F 1,49 = 0 15 P = 0 6988 Litter size F 1,49 = 1 19 P = 0 2817 Experimentally induced Treatment F 1,55 = 7 78 P = 0 0074 Birth date F 1,55 = 4 16 P = 0 0330 Treatment Birth date F 1,55 = 7 79 P = 0 0074 Litter size F 1,55 = 0 32 P = 0.5759 Experimentally prolonged Treatment F 1,25 = 14 77 P = 0 0008 Litter size F 1,25 = 0 85 P = 0 3666 to the 36 females in the synchronous (experimentally induced) treatment and 86 offspring born to the 46 females in the asynchronous (naturally born) treatment. Of these, 17 offspring were removed from the analysis (because they refused to run, attacked the brush, or ran in the wrong direction) and 12 were excluded because they came from litters with single offspring resulting in a sample of 127 offspring (Table 1). There were significant differences between treatments in average sprint speed (Table 2), with offspring from the asynchronous treatments significantly faster than those from the synchronous treatment. However, there were no overall differences in average locomotor performance between first and last born offspring in either treatment (Table 2). Across treatments, both date of birth (Table 2) and mass at birth (Table 2) were the strongest predictors of offspring locomotor performance, but in both cases this relationship was weak (r 2 = 0 10 and 0 03 respectively). There was a significant interaction between treatment and litter date of birth on the within litter locomotor difference (Table 3). Closer examination within treatments revealed that where sibling birth date was identical (experimentally induced treatment) there was no shift throughout the season in the within litter locomotor difference, but where birth of second and third offspring was delayed (naturally born treatment) there was a significant negative shift. The direction of this effect was the same as that of the natural population, with early born litters exhibiting a positive difference in locomotor performance between first and last born offspring (i.e. first born offspring were faster than last born offspring), whereas litters born later in the season had a negative litter locomotor difference on average (i.e last born offspring were faster than first born offspring) (Fig. 2). EXPERIMENTAL PROLONGATION OF PARTURITION Of the 44 females brought into the laboratory to give birth, 38 did so, resulting in 80 offspring (Table 1). This was comprised of 21 females in the long basking treatment that gave birth to 46 offspring and 17 females in the short basking treatment that gave birth to 34 offspring. Of these, 4 offspring were removed from the analysis because they came from single Fig. 2. Data from the induction of birth experiment showing the interaction between treatment and date of birth on the within litter locomotor difference between first and last born offspring. The graph shows the significant negative relationship (solid line) between date of birth and the within litter difference in sprint speed (m s 1 ) for asynchronously born offspring ( ) and the lack of a relationship (dashed line) for synchronously born offspring ( ). The dotted line represents a difference of 0 where both offspring have an equal average sprint speed. Above the line, first born offspring have a faster average sprint speed, below the line, second born offspring have a faster average sprint speed. offspring litters, resulting in a total sample of 76 offspring. We found a significant overall effect of basking regime on offspring locomotor performance; offspring from females with extended opportunity to bask had significantly faster sprint speeds than those from females under reduced basking regime (Table 2). Basking regime also had a significant effect on the relative locomotor performance of first vs. last born offspring, with first born offspring faster than last born in the extended basking treatment and last born faster than first born offspring in the reduced basking treatment (treatment birth order interaction; Fig. 3). These patterns were supported by a treatment effect on the within litter locomotor difference, with a small positive difference in the long basking treatment (i.e. first born offspring faster) and a strong negative difference in the short basking treatment (i.e. last born offspring faster) (Table 3). There were also significant differences between basking regimes in gestation length, and therefore offspring birth date (G. M. While, unpublished data). Thus, when we

Delayed parturition and locomotor performance 823 Fig. 3. Difference in the sprint speed of first ( ) and last ( ) born Egernia whitii offspring born to those females held under long basking conditions during gestation and those females held under short basking conditions during gestation. compare these results to those from the two previous data sets, this suggests a seasonal shift in the relative performance of first vs. last born offspring. Discussion Our data represent the first experimental test of the hypothesis that prolonged pregnancy beyond that required to complete development is favoured through increased performance of hatchlings. This supports correlative data from other viviparous lizards which has shown that performance (sprint speed or stamina) increases with birth date (Shine & Harlow 1993; Shine & Downes 1999; Shine & Olsson 2003). There are several explanations for this, the most parsimonious being that female (or offspring) traits covary with timing of ovulation and maternal basking, which together are the most important predictors of birth date in squamates (Beuchat 1988; Shine & Harlow 1993; Wapstra 2000). However, a similar result was also found in a lizard in which offspring are fully developed at the onset of hibernation and born later in spring, which is not as easily explained (Shine & Olsson 2003). Shine and Olsson (2003) suggested that prolonging pregnancy, beyond that required for complete development, in viviparous reptiles and other species could be an adaptive strategy to enhance offspring performance through increased prenatal development of musculoskeletal or brain function. Our correlative and experimental results show mixed support for the hypothesis that facultative retention of fully formed offspring has direct beneficial effects on offspring performance subsequent to parturition in E. whitii. Although experimental induction of birth showed that differences in locomotor performance between offspring were, at least partly, driven by extended retention of embryos, we found no evidence that it enhanced the locomotor performance of last born offspring. On the contrary, in the natural population there was evidence that the opposite may be true, with last born offspring within a litter exhibiting a poorer locomotor performance than first born offspring. However, this pattern depends to some extent on the thermal conditions experienced during gestation and thus birth date. This was shown by the significant variation in the relative performance of first and last born offspring (the within litter locomotor difference) across birth dates, with first born offspring faster primarily early in the season, but last born offspring faster primarily later in the season. While these patterns were relatively weak within the natural population, when we tested this by experimentally manipulating offspring birth date the above patterns strengthened. Whereas females provided with good basking conditions from early gestation in the laboratory (and gave birth early in the season) had significantly faster first born offspring, females with poor basking conditions (and gave birth later in the season and at the extreme end of the birth date distribution in the natural population), had significantly faster last born offspring. In other words, the relative difference between first and last born offspring was reversed between treatments. Thus, differences in the strength and direction of the effect of birth date on the relative performance across the three data sets is therefore most parsimoniously explained by differences in birth date per se. We discuss the implications of these results for the evolution of maternal effects on performance of offspring and birthing asynchrony. Combined, these results suggest that there are two, independent, effects on offspring locomotor performance that are linked to pregnancy duration E. whitii. First, locomotor performance is generally negatively affected by an increase in gestation length. Although this could be driven by a number of factors in natural populations, our and others experimental data suggest that one of the most important is variation in thermal microhabitat (via its effect on embryonic development, Elphick & Shine 1998; Qualls & Andrews 1999). Second, facultative retention of fully formed offspring can result in both an increase and a decrease of offspring locomotor performance (i.e. the effect is dependent on date of birth). Thus, rather than having a consistent positive or negative effect, our results indicate a strong seasonal shift in both absolute and relative locomotor performance of offspring from different birth order positions, suggesting that maternal environmental conditions can mediate both the average and the relative within-clutch performance of juveniles. These seasonal shifts in the relative performance of offspring from different birth order positions may be crucial in understanding the birthing asynchrony patterns observed in this genus. Despite being widespread within the lineage (Chapple 2003), our understanding of the underlying reasons for birthing asynchrony is still relatively poor. Recent evidence suggests birthing asynchrony increases asymmetries in sibling competition (While et al. 2007; While & Wapstra 2008), which may facilitate the establishment of the social groups which are characteristic of this and other Egernia species (e.g. Gardner et al. 2001; O Connor & Shine 2003; Chapple & Keogh 2006). The observed effects of birthing asynchrony on offspring performance is in line with this argument and could provide a means by which mothers manipulate sibling conflict and the relative competitive ability or dispersal propensity of first vs. last born offspring according to environmental conditions (and thus offspring birth date). This may be important if the advantages of asynchrony differ depending

824 G. M. While et al. on environmental conditions (Wiebe 1995) and could ensure flexibility in the degree of sibling hierarchy in the face of fluctuating selection. However, although several studies document positive selection on locomotor performance (Jayne & Bennett 1990; Warner & Andrews 2002; Miles 2004; Husak 2006; Husak et al. 2006; Irschick & Meyers 2007), it is currently unknown to what extent variation in locomotor performance captures variation in traits that affect withinlitter competition for access to parental resources or the propensity to disperse. Furthermore, selection on locomotor performance is expected to differ depending on environmental conditions and other offspring traits, including those affected by maternal effects. For example, in common lizards Lacerta vivipara, the fitness benefits of stamina at birth depend on juvenile food conditions (Le Galliard et al. 2004). At low food availability, high-performing hatchlings had higher survival than low-performing hatchlings. At high food availability, however, the patterns disappeared. Thus, the relative competitive ability of faster vs. slower offspring, and therefore the outcome of sibling differences in sprint speed, may differ across birth dates in E. whittii. Even consistent selection for creation of sibling hierarchies or parental favouritism could therefore result in seasonal shifts in performance of juveniles. Three points suggest caution with the above conclusions. First, we know very little of the means by which females could facultatively alter the relative performance of offspring. Previous research suggests that prolonged retention of offspring may influence offspring locomotor performance through enhanced prenatal development of musculoskeletal or brain function (Shine & Olsson 2003, see also Vince & Chinn 1971; Woolf et al. 1976). However, given the within season shift in relative locomotor performance of first and last born offspring in this system, this explanation is probably unlikely. Alternatively, females may alter offspring performance ability through a number of other mechanisms, such as altering order specific resource allocation at ovulation (e.g. Schwabl et al. 1997; Badyaev et al. 2006), and several may be employed simultaneously. While this remains to be explored, evidence from the hormonal induction of birth experiment, where induction resulted in, on average, a reduction in the difference between first and last born offspring both early and late in the season, suggests that facultative retention of offspring does play some part in the within litter differences in offspring locomotor performance. Second, as noted above, our understanding of the links between performance traits, such as locomotor performance, and fitness related traits, such as competitive ability, dispersal, and survival of offspring need to be improved. For example, while locomotor performance in Egernia may be an important predictor of offspring dispersal (and thus impact on social spacing) or the ability of offspring to avoid predators and/or infanticidal conspecifics, we know little of the functional contexts within which enhanced locomotor performance is favoured. Third, the extent to which birthing asynchrony and within-clutch variation in offspring traits enhance or reduce conflicts between siblings in Egernia is poorly understood. From a theoretical perspective, maternal manipulation of offspring birth order according to the maternal, but not offspring, optima could be evolutionarily unstable as evolution of offspring counter responses should occur (Uller 2008). Indeed, there is some evidence that the offspring, rather than the mother, induce parturition in mammals (Johnson & Everitt 1988; Shaw & Renfree 2001) and there is the potential for similar mechanisms to occur in squamates (Girling & Jones 2006). Studies of proximate mechanisms of induced parturition in this and other viviparous non-mammalian vertebrates are currently lacking (but see Girling & Jones 2006), but could be important for understanding the evolutionary dynamics of these systems (Uller 2008). In summary, we show that facultative retention of offspring in the oviduct can have both positive and negative effects on offspring sprint speed relative to birth order position, resulting in a reversal of the relative sprint speed of first and last born offspring across a season. Thus, although prolonging pregnancy in skinks does not have consistent positive effects on physiological performance as previously suggested (Shine & Olsson 2003), it could be important for facultative adjustment of the strength and direction of sibling hierarchies in this and other species. Acknowledgements We are grateful for the detailed and constructive comments and suggestions made by two anonymous referees. This work was partially funded by the Holsworth Wildlife Research Fund, the Environmental Futures Network, the Association for the Study of Animal Behaviour (to GMW), the Wenner-Gren Foundations (to TU), and the Australian Research Council (to TU and EW). All work complied with wildlife regulations imposed by the Tasmanian Department of Primary Industries and Water (Permit No. FA 06538) and the Animal Ethics Committee at the University of Tasmania (Protocol No. A0008524; A0008055). References Atkins, N., Jones, S.M. & Guillette, L.J. Jr (2006) Timing of parturition in two species of viviparous lizard: influences of β-adrenergic stimulation and temperature upon uterine responses to arginine vasotocin (AVT). Journal of Comparative Physiology, 176, 783 792. Badyaev, A.V., Acevedo, S.D., Navara, K.J., Hill, G.E., & Mendonca, M.Y. (2006) Evolution of sex-biased maternal effects in birds: III. Adjustment of ovulation order can enable sex-specific allocation of hormones, carotenoids, and vitamins. Journal of Evolutionary Biology, 19, 1044 1057. Beuchat, C.A. (1988) Temperature effects during gestation in a viviparous lizard. Journal of Thermal Biology, 13, 135 142. Chapple, D.G. (2003) Ecology, life-history, and behaviour in the Australian Scincid genus Egernia, with comments on the evolution of complex sociality in lizards. Herpetological Monographs, 17, 145 180. Chapple, D.G. & Keogh, J.S. (2006) Group structure and stability in social aggregations of White s skink, Egernia whitii. Ethology, 112, 247 257. Cree, A. & Guillette, L.J. Jr (1991) Effect of β-adrenergic stimulation on uterine contraction in response to arginine vasotocin and prostaglandin-f2-α in the gecko Hoplodactylus maculatus. Biology of Reproduction, 44, 499 510. Deeming, D.C. (2002) Avian incubation: behaviour, environment and evolution. Oxford University Press, Oxford. Elphick, M.J. & Shine, R. (1998) Long term effects of incubation temperatures on the morphology and locomotor performance of hatchling lizards (Bassiana duperreyi, Scincidae). Biological Journal of the Linnean Society, 63, 429 447. Gardner, M.G., Bull, C.M., Cooper, S.J.B. & Duffield, G.A. (2001) Genetic evidence for a family structure in stable social aggregations of the Australian lizard Egernia stokesii. Molecular Ecology, 10, 175 183. Girling, J.E. & Jones, S.M. (2006) In vitro steroid production by adrenals and kidney-gonads from embryonic southern snow skinks (Niveoscinscus microlepidotus): implications for the control of the timing of parturition? General and Comparative Endocrinology, 145, 169 176.

Delayed parturition and locomotor performance 825 Husak, J.F. (2006) Does speed help you survive? A test with collard lizards of different ages. Functional Ecology, 20, 174 179. Husak, J.F., Fox, S.F., Lovern, M.B. & Van den Bussche, R.A. (2006) Faster lizards sire more offspring: sexual selection on whole-animal performance. Evolution, 60, 2122 2130. Irschick, D.J. & Meyers, J.J. (2007) An analysis of the relative roles of plasticity and natural selection on morphology and performance in a lizard (Urosaurus ornatus). Oecologia, 153, 489 499. Jayne, B.C. & Bennett, A.F. (1990) Selection on locomotor performance capacity in a natural population of garter snakes. Evolution, 44, 1204 1229. Johnson, M. & Everitt, B. (1988) Essential reproduction. 3rd ed. Blackwell Scientific Publications, Oxford. Le Galliard, J-F., Clobert, J. & Ferriere, R. (2004) Physical performance and darwinian fitness in lizards. Nature, 432, 502 505. Littell, R.C., Milliken, G.A., Stoup, W.W., Wolfinger, R.D. & Schabenberger, O. (1996) SAS for Mixed Models. SAS Institute, Cary. Losos, J.B., Creer, D.A. & Schulte, J.A. (2002) Cautionary comments on the measurement of maximum locomotor capabilities. Journal of Zoology, 258, 57 61. Meylan, S. & Clobert, J. (2004) Maternal effects on offspring locomotion: influence of density and corticosterone elevation in the lizard Lacerta vivipara. Physiological and Biochemical Zoology, 77, 450 458. Miles, D.B. (2004) The race goes to the swift: fitness consequences of variation in sprint performance in juvenile lizards. Evolutionary Ecology Research, 6, 63 75. Mousseau, T.A. & Fox, C.W. (1998) Maternal effects as adaptations. Oxford University Press, Oxford. O Connor, D. & Shine, R. (2003) Lizards in nuclear families : a novel reptilian social system in Egernia saxatilis (Scincidae). Molecular Ecology, 12, 743 752. Olsson, M. & Shine, R. (1998) Timing of parturition as a maternal care tactic in an alpine lizard species. Evolution, 52, 1861 1864. Olsson, M. & Shine, R. (2003) Female-biased natal and breeding dispersal in an alpine lizard, Niveoscincus microlepidotus. Biological Journal of the Linnean Society, 79, 277 283. Qualls, C.P. & Andrews, R.M. (1999) Cold climates and the evolution of viviparity in reptiles: cold incubation temperatures produce poor-quality offspring in the lizard, Schloporus virgatus. Biological Journal of the Linnean Society, 67, 353 376. Schwabl, H., Mock, H.W. & Gieg, J.A. (1997) A hormonal mechanism for parental favouritism. Nature, 386, 231 231. Shaw, G. & Renfree, M.B. (2001) Fetal control of parturition in marsupials. Reproduction Fertility and Development, 13, 653 659. Shine, R. & Downes, S.J. (1999) Can pregnant lizards adjust their offspring phenotypes to environmental conditions? Oecologia, 119, 1 8. Shine, R. & Harlow, P. (1993) Maternal thermoregulation influences offspring viability in a viviparous lizard. Oecologia, 96, 122 127. Shine, R. & Olsson, M. (2003) When to be born? Prolonged pregnancy or incubation enhances locomotor performance in neonatal lizards (Scincidae). Journal of Evolutionary Biology, 16, 823 832. Sinn, D.L., While, G.M. & Wapstra, E. (2008) Maternal care in a social lizard: links between female aggression and offspring fitness. Animal Behaviour, 76, 1249 1257. Sorci, G. & Clobert, J. (1997) Environmental maternal effects on locomotor performance in the common lizard (Lacerta vivipara). Evolutionary Ecology, 11, 531 541. Swain, R. & Jones, S.M. (2000) Maternal effects associated with gestation conditions in a viviparous lizard, Niveoscincus metallicus. Herpetological Monographs, 14, 432 440. Uller, T. (2008) Developmental plasticity and the evolution of parental effects. Trends in Ecology and Evolution, 23, 432 438. Uller, T. & Olsson, M. (2003) Prenatal exposure to testosterone increases ectoparasite susceptibility in the common lizard, Lacerta vivipara. Proceedings of the Royal Society of London B, 270, 1867 1870. Uller, T & Olsson, M. (2006) Direct exposure to corticosterone during embryonic development influences behaviour in an ovoviviparous lizard. Ethology, 112, 390 397. Vince, M.A. & Chinn, S. (1971) Effect of accelerated hatching on the initiation of standing and walking in the Japanese quail. Animal Behaviour, 19, 62 66. Walker, J.A., Ghalambor, C.K., Griset, O.L., McKenney, D. & Reznick, D.N. (2005) Do faster starts increase the probability of evading predators? Functional Ecology, 19, 808 815. Wapstra, E. (2000) Maternal basking opportunity affects juvenile phenotype in a viviparous lizard. Functional Ecology, 14, 345 352. Wapstra, E., Swain, R., Jones, S.M. & O Reilly, J. (1999) Geographical and annual variation in reproductive cycles in the Tasmanian spotted snow skink, Niveoscincus ocellatus (Squamata: Scincidae). Australian Journal of Zoology, 47, 539 550. Warner, D.A. & Andrews, R.M. (2002) Laboratory and field experiments identify sources of variation in phenotypes and survival of hatchling lizards. Biological Journal of the Linnean Society, 76, 105 124. Warner, D.A. & Shine, R. (2006) Morphological variation does not influence locomotor performance within a cohort of hatchling lizards (Amphibolurus muricatus, Agamidae). Oikos, 114, 126 134. Watkins, T.B. (1996) Predator-mediated selection on burst swimming performance in tadpoles of the Pacific tree frog, Pseudacris regilla. Physiological Zoology, 69, 154 167. Weekes, H.C. (1935) A review of placentation among reptiles. Proceedings of the Royal Society of London, Series B, 3, 625 645. While, G.M. & Wapstra, E. (2008) Are there benefits to being born asynchronously: an experimental test in a social lizard. Behavioral Ecology, 19, 208 216. While, G.M., Jones, S.M. & Wapstra, E. (2007) Birthing asynchrony is not a consequence of asynchronous offspring development in a non-avian vertebrate, the Australian skink Egernia whitii. Functional Ecology, 21, 513 519. While, G.M., Uller, T. & Wapstra, E. (2009) Within-population variation in social strategies characterise the social and mating system of an Australian lizard, Egernia whitii. Austral Ecology, in press. Wiebe, K.L. (1995) Intraspecific variation in hatching asynchrony: should birds manipulate hatching spans according to food supply? Oikos, 74, 453 462. Woolf, N.K., Bixby, J.L. & Capranica, R.R. (1976) Prenatal experience and avian development: brief auditory stimulation accelerates the hatching of Japanese quail. Science, 194, 959 960. Received 12 June 2008; accepted 20 January 2009 Handling Editor: David Reznick