Effects of nest temperature and moisture on phenotypic traits of hatchling snakes (Tropidonophis mairii, Colubridae) from tropical Australia

Transcription

1 Blackwell Publishing LtdOxford, UKBIJBiological Journal of the Linnean Society24-466The Linnean Society of London, 26? Original Article INCUBATION EFFECTS IN A SNAKE G. P. BROWN and R. SHINE Biological Journal of the Linnean Society, 26, 89, With 5 figures Effects of nest temperature and moisture on phenotypic traits of hatchling snakes (Tropidonophis mairii, Colubridae) from tropical Australia GREGORY P. BROWN and RICHARD SHINE* Biological Sciences A8, University of Sydney, NSW 26, Australia Received 1 January 25; accepted for publication 4 January 26 Previous research on developmentally plastic responses by reptile embryos has paid relatively little attention to tropical species, or to possible interactions between the effects of thermal and hydric regimes. In the present study, eggs of keelback snakes (Tropidonophis mairii), from a tropical area with strong temporal and spatial variation in soil temperatures and moisture levels, were incubated. The phenotypic traits of hatchling snakes (body size, shape, muscular strength) were affected by moisture content of the incubation medium (vermiculite plus 1% vs. 5% water by mass), by mean incubation temperatures (25.7 vs C) and by diel thermal variation (diel range 6. vs. 8.4 C). Interactions between these factors were negligible. Cooler, more thermostable, moister conditions resulted in larger offspring, a trait under strong selection in this population. Thermal and hydric conditions covary in potential nestsites (e.g. deeper nests are more thermostable as well as moister). This covariation may influence the evolution of reaction norms for embryogenesis. For example, if moister nests enhance offspring fitness and are cooler, then selection will favour the ability to develop in cool as well as moist conditions. Thus, the evolution of optimal incubation conditions with respect to one variable (e.g. temperature) may be driven by patterns of association with another variable (e.g. soil moisture) among natural nest-sites. Perhaps for this reason, the thermal optimum for incubation is surprisingly low in this tropical species. 26 The Linnean Society of London, Biological Journal of the Linnean Society, 26, 89, ADDITIONAL KEYWORDS: developmental plasticity hydric incubation reptile thermal. INTRODUCTION Developmental plasticity is widespread, almost ubiquitous (Sultan, 1987; Via et al., 1995). In organisms as diverse as trees and mammals, pathways in early development can be modified by the exposure of the embryo to external factors such as temperature, moisture, or nutrient availability. In turn, resultant effects on the phenotypes of offspring potentially can alter organismal fitness, and thus play a critical role in generating the phenotypic variance that provides the raw material for natural selection (Seigel & Ford, 1991; Rhen & Lang, 1995; Via et al., 1995; Sinervo & Svensson, 1998). Information on the nature and degree of developmental plasticity is essential before we can understand issues such as the extent to which natural *Corresponding author. rics@bio.usyd.edu.au selection on phenotypic variation among individuals within a population will modify frequencies of genes that code for specific traits, genes that code for norms of reaction of those traits, or genes that code for parental behaviours that influence the early nest environment (Via & Lande, 1985; Scheiner & Callahan, 1999; Shine, 24). Unfortunately, the challenge is a formidable one because natural incubation environments typically differ in multiple attributes (e.g. temperature, moisture), with both means and variances of such abiotic factors often varying through space and time (Muth, 198; Packard, Miller & Packard, 1993). For example, moister nests often will be cooler, and a shaded nest will have lower diel thermal variance as well as a lower mean temperature. Thus multifactorial experimental approaches are required to identify possible interactions among factors, and the effects of variation, as well as mean values for abiotic factors, need to 159

2 16 G. P. BROWN and R. SHINE be considered (Ballinger, 1983; Ji & Brana, 1999; Flatt et al., 21). Inevitably, research on the evolutionary and ecological implications of developmental plasticity has tended to focus on a limited number of abiotic factors, on some types of organisms rather than others, and on organisms in particular kinds of environments. Such biases are readily apparent in studies of developmental plasticity in reptiles, as is evident in recent reviews of this field (Birchard, 24; Deeming, 24; Shine, 24). For example, interest in the phenomenon of temperature-dependent sex determination has generated extensive research on developmental responses to incubation temperature in turtles and crocodilians whereas studies investigating responses to the hydric environment are less popular. Despite their much higher species diversity, squamate reptiles have attracted less attention; this is particularly true for snakes. Lastly, the philopatry of most professional scientists has resulted in a scarcity of research on tropical as opposed to temperate-zone reptiles. In an attempt to counter such biases, we have conducted detailed ecological studies on developmental plasticity in a tropical snake. We have already published the results of previous experimental studies on the phenotypic responses of this species to incubation conditions, notably the magnitude of diel variation in nest temperatures (Webb, Brown & Shine, 21), the mean water content of the nest substrate (Shine & Brown, 22; Brown & Shine, 24), and gradual declines or increases in soil moisture content over the course of incubation (Brown & Shine, 25). In the present study, the previous work is extended to explore the roles of temperature and moisture simultaneously rather than independently (i.e. with a factorial design that allows examination of interactive effects between moisture and temperature). Additionally, the effects of mean incubation temperature, as well as diel thermal variance, are considered. MATERIAL AND METHODS STUDY AREA AND SPECIES The study took place in the Fogg Dam Nature Reserve, 6 km south east of Darwin, in Australia s Northern Territory. The Fogg Dam wall is a 1.3-km long embankment constructed on the Adelaide River floodplain. The area south of the wall contains permanent water and the north side consists of seasonally inundated floodplain. The floodplain is usually, but is shallowly inundated by monsoonal rains for up to 5 months each year (Madsen & Shine, 1996). By contrast to the highly seasonal precipitation regime, temperatures remain warm year-round (monthly mean maxima C, minima C). A variety of soil types occurs in the region, with deeply cracking black soils on the floodplain and sandier red soil at higher, drier elevations (Madsen & Shine, 1996). The study area has been described and illustrated in detail elsewhere (Madsen & Shine, 1996; Brown & Shine, 25). Keelbacks (Tropidonophis mairii) are natricine colubrids (up to.8 m in length) that feed primarily on frogs (Shine, 1991) and are widely distributed in nearcoastal habitats throughout much of tropical and subtropical Australia (Cogger, 1992). In our study area, female keelbacks produce multiple clutches within a single breeding season, with oviposition from April through November each year (Brown & Shine, 22). Thus, keelback eggs laid at different times of year or in different soil types are likely to experience very different conditions in terms of both the thermal and hydric environment. Thermal fluctuations during incubation strongly influence the phenotypic traits of hatchling keelbacks (Webb et al., 21). Substrate water content also significantly modifies the phenotypic traits of hatchlings, either when eggs are kept throughout incubation under vs. conditions (Shine & Brown, 22; Brown & Shine, 24) or when the eggs experience progressively ter or drier conditions through the course of incubation (Brown & Shine, 25). The present study explores the effects of mean incubation temperature, and possible interactions between thermal and hydric regimes. PROCEDURES To quantify thermal profiles within keelback nests and potential nests-sites, soil temperatures were monitored at 2-h intervals: (1) over 58 days within four natural nests (these nests were located by finding females in the process of ovipositing; on excavation, one of these nests actually contained an additional, earlier clutch of keelback eggs and ten small lizard eggs) and (2) by placing data-loggers at depths of, 1, 2, and 3 cm at two sites along the dam wall; one in a shaded area where paperbark (Melaleuca) trees line the wall on the eastern edge of the dam, and another more exposed area midway along the dam wall. Many gravid female keelbacks are captured in both areas during the nesting season. These potential nest-sites were monitored from May 22 to July 23. To quantify hydric regimes, soil samples were taken at, 5, and 1 cm below the soil surface at the same sites over the same period, and these samples were dried to constant weight in the laboratory, as previously described (Shine & Brown, 22). Gravid keelbacks collected during nightly surveys of the dam wall were returned to the field laboratory where they were maintained in captivity until oviposition. Females were kept separately in plastic boxes (4 3 2 cm) lined with newspaper. Each cage

3 INCUBATION EFFECTS IN A SNAKE 161 included a water dish and a shelter/nest-site consisting of a 29-mL black plastic dish containing damp vermiculite. Oviposition occurred 4 8 days after capture. Females were released at their capture sites after oviposition. Eggs were weighed, and then placed in individual incubation vials (6.5 cm diameter 4.5 cm high) containing an amount of vermiculite equal to twice the egg s mass. These vials were then allocated to experimental incubation treatments that differed in hydric and thermal conditions, in a splitclutch design. Half the eggs were incubated in vermiculite (5% water by weight) whereas the others were kept moister (1% water by weight). Orthogonal to this hydric treatment, eggs were placed in one of three thermal treatments: (1) low mean temperature, low diel variance; (2) low mean temperature, high diel variance; and (3) high mean temperature, low diel variance. Ideally, one further treatment (high mean temperature, high diel variance) would have been included to allow fully factorial analysis, but the clutch sizes were too low to permit splitting most clutches into so many replicates. Thus, the design adopted allowed two separate analyses to be conducted with respect to temperature (and its interaction with moisture): (1) the effect of diel thermal variance (i.e. treatment 1 vs. 2) and (2) the effect of mean temperature (i.e. treatment 1 vs. 3). Thus, two separate two-factor analyses were conducted, one looking at eggs incubated in vs. conditions under high vs. low diel thermal variance, and the other looking at eggs incubated in vs. conditions under high vs. low mean temperatures. Tweleve eggs were used from each of four clutches (total = 48 eggs), with four eggs from each clutch being allocated to each of the three temperature regimes; two of them on moist vermiculite and the other two on vermiculite. Temperature-controlled incubators were not available, and so less precise means of generating the various thermal conditions required were used. Eggs in the high-thermal-variance temperature treatment were kept in an uninsulated plastic box in a room within the field laboratory; air-conditioning maintained the temperature within this room at C. The low-thermal-variance eggs were kept in the same room, but inside an insulated styrofoam cooler. The high-mean-temperature treatment was achieved by placing the eggs in an insulated cooler on top of a water heater in an adjacent room. Thermal data-loggers were placed within incubation containers to quantify the temperatures actually experienced by eggs under these conditions. Eggs were weighed weekly and water was replaced in treatment containers to replace evaporative loss. No water was replaced in the treatment containers; thus, hydric variance, as well as mean moisture content, may have differed between these two treatments. The standard morphometric traits of hatchling snakes were recorded within 24 h of emergence from the egg, and the muscular strength of hatchlings was measured at 1 week of age before releasing the young snakes at the site of their mother s capture. Typically, studies on reptiles use locomotor speed as a measure of organismal performance. The estimate of muscular strength in the present study is highly correlated with mean swim speed over 1.5 m (N = 411, r =.62, P <.1; G. P. Brown, unpubl. data), and may provide a more general measure of performance because it relates to a hatchling s ability to subdue prey as well as locomotor ability. Strength was measured by affixing the snake s head to the top of a table with electrical tape across the neck. The clamp of a 5-g Pesola scale was padded with foam rubber and clipped to the base of the hatchling s tail. The scale was positioned posterior to the snake and in line with its body, and the snake was then induced to contract its body by gently touching its head. The maximum force of the contraction was registered by the scale. Each snake was induced to pull on the scale seven times, with successive pulls at 2 3 s intervals. Trials were conducted at C (for further details, see Shine & Brown, 22; Brown & Shine, 24). RESULTS SOIL TEMPERATURES Vertical thermal stratification at potential nest sites was limited, with temperatures at the shaded soil surface generally similar to those 3 cm below (Fig. 1). However, soil temperatures in the shady (tree-lined) section of the dam wall averaged approximately 6 C lower than those measured in a site exposed to full sun (Fig. 1; means = 27.9 vs C). The effect of shade on soil temperatures was also evident among the four natural nests. Two nests located at the base of trees remained cool, averaging 23. and 23.4 C and with mean diel variances of.8 and 3.3 C, respectively (Fig. 2). The two nests (including the communal nest) located on the unshaded section of the dam were hotter (3.6 C and 29. C) but with similar diel ranges (1.8 C and 2.8 C). Measurements of soil temperatures and water content showed that mean temperatures remained constant for at least the top 3 cm of the soil profile, but diel variance decreased rapidly at greater depth whereas soil water content increased (Fig. 3). Thus, nesting female keelbacks not only have abundant opportunity to select nest-sites with different mean temperatures (by choosing sites in sun vs. shade, above), but also have access to sites that differ in diel thermal variation and hydric conditions (by selecting the depth of nests). However, the latter two

4 162 G. P. BROWN and R. SHINE 4 Soil temperature ( C) Week Figure 1. Mean weekly soil temperature profiles at Fogg Dam from May 22 to July 23. Solid lines are shaded surface temperatures, dashed lines are temperatures 3 cm below the soil surface. The upper two lines are from an open area midway along the dam wall, whereas the lower set of lines is from a shaded area in a tree-lined corridor near the eastern end of the dam wall. The open circles depict mean weekly soil temperatures of four natural keelback nests: one monitored in 21, one in 23, and two in nest hot low high % frequency Temperature ( C) Figure 2. Comparison between mean thermal regimes measured in four natural keelback nests vs. those used for experimental incubation of keelback eggs in the present study. The lines show the mean proportions of observations in each 1 C interval, with data for the four natural nests combined.

5 INCUBATION EFFECTS IN A SNAKE 163 Temperature ( C) mean temperature mean diel thermal range soil water content Soil moisture (%) COMPARISON BETWEEN INCUBATION REGIMES AND TEMPERATURES IN NATURAL NESTS Figure 2 compares the temperatures measured within our incubation groups with those that were recorded in natural nests at the study site. The mean temperatures in the experimental treatments were intermediate between the two cool and two warm natural nests. Levels of variation in temperatures of experimental treatments were within the thermal ranges measured in natural nests (Fig. 2). Thermal regimes in potential nest-sites were generally hotter and more variable than in the experiments (Fig. 1). Thus, the experimental thermal regimes offer reasonable simulations of the kinds of nest conditions used by, and available to, nesting keelbacks Depth (cm) Figure 3. Changes in thermal and hydric conditions as a function of depth beneath the soil surface for potential nesting sites during the keelback nesting season (May to November) at Fogg Dam. The graph plots values (mean ± standard error) for the mean temperature, the mean diel temperature range, and soil water content. Soil moisture was measured only at depths of (immediately below the soil surface), 5, and 1 cm, whereas temperatures were recorded at 1, 2, and 3 cm. These figures are based on 298 days of sampling daily temperature, and 19 readings of soil moisture levels. variables are linked, in that a deeper nest will be moister, as well as having a lower diel thermal variance (Fig. 3). LABORATORY INCUBATION REGIMES Mean temperatures experienced by eggs over the course of incubation differed significantly between our three thermal treatments [one-factor analysis of variance (ANOVA) with four replicates per treatment (one per litter), F 2,9 = 123.7, P <.1]. Post-hoc tests showed that mean temperatures of the low mean, high variance and low mean, low variance treatments did not differ significantly from each other (25.4 vs C, respectively, P >.5) but both were lower than the high mean, low variance treatment (27.9 C; P <.5). Diel thermal variances were higher for the low mean, high variance eggs (mean diel range 8.4 C) than for either of the low variance treatments (5.6 and 5.8 C, respectively; F 2,336 = 15.8, P <.1). Thus, the makeshift incubators used succeeded in generating thermal regimes conforming to the aims of the study. EFFECTS OF WATER AVAILABILITY AND MEAN INCUBATION TEMPERATURE Figure 4 shows effects of incubation regimes on egg and hatchling phenotypes, and Table 1 presents the results from two-way ANOVAs performed on these data. These analyses show that both mean temperature and substrate moisture affected most hatchling traits that were measured, as well as the mass of the egg prior to hatching (Table 1). Mean temperature also affected the incubation period. To assess the statistical significance of these patterns without the problem of artefactual data resulting from multiple testing of related data sets, a multivariate ANOVA (MANOVA) was performed on the effects of the experimental treatments on the phenotypic traits of hatchlings [snout vent length (SVL), head length, tail length, mass, maximum strength, mean strength and SD of strength]. This MANOVA confirmed that mean temperature and moisture both had significant effects on hatchling phenotype, but there was no significant interaction between these factors (temperature, F 1,29 = 3.9, P <.6; moisture, F 1,29 = 6.5, P <.3; interaction, F 1,29 = 1.2, P =.34). Most of the hatchling phenotypic variables measured (head length, tail length, mass and strength) depend upon hatchling body length (SVL). Thus, the effects of incubation treatment on these variables (Table 1) might be epiphenomena: secondary consequences of incubation-induced effects on overall body size. To evaluate this possibility, ANCOVAs were conducted with hatchling SVL as the covariate (i.e. to assess effects of temperature and moisture regimes on hatchling phenotypes when correcting for hatchling SVL). Relative to SVL, hatchling body mass was the only variable affected by incubation regimes (Table 2). The overall multivariate analysis of covariance (MANCOVA) confirmed that most of the moisture and temperature effects on hatchling phenotypes were due

6 164 G. P. BROWN and R. SHINE 16 A.1 B cold Snout-vent length (cm) Body condition hot C.8 D Relative head length Relative strength Incubation moisture Incubation moisture Figure 4. Mean values and associated standard errors for phenotypic traits of hatchling keelback snakes incubated under different conditions. The traits are snout vent length (A), body condition (B), relative head length (C), and relative strength (D). Body condition and relative traits are based on residual scores from linear regressions of the trait in question (ln body mass, head length, or mean strength) against ln snout vent length. The eggs were incubated at different mean temperatures ( cold 25.7 vs. hot 27.9 C) and in vermiculite of different substrate moisture contents ( 5% vs. 1% water by weight). Vertical bars indicate standard errors. to differences in hatchling SVL between incubation regimes. After correcting for SVL, moisture had a significant effect on overall hatchling phenotypes (F 1,29 = 3.2, P <.2) but temperature did not (F 1,29 = 1.9, P =.12; interaction, F 1,29 =.9, P =.51). EFFECTS OF WATER AVAILABILITY AND DIEL VARIANCE IN INCUBATION TEMPERATURE ANOVAs revealed that, as for the analyses above, water content of the incubation substrate significantly affected most phenotypic traits that were measured on hatchling keelbacks (Table 3; Fig. 5). Diel thermal variance significantly modified hatchling tail length and muscular strength, but no other traits (Table 3). A MANOVA showed that, overall, the phenotypic traits of hatchling snakes were affected significantly by thermal variance (F 1,29 = 5.3, P <.2) and moisture (F 1,29 = 6.9, P <.3) but not by any interaction between these factors (F 1,29 = 1.2, P =.35). As for the preceding comparison, some of these effects might be due to incubation-induced differences in overall hatchling size. Thus, hatchling SVL was incorporated as a covariate and ANCOVAs and a MANCOVA were conducted on these data. Some previously significant effects disappeared but others

7 INCUBATION EFFECTS IN A SNAKE 165 Table 1. Effects of incubation conditions (mean nest temperature and substrate water content) on phenotypic traits of eggs and hatchlings of keelback snakes Mean temperature (warm vs. cool) Moisture content ( vs. ) Interaction Variable F 1,29 P F 1,29 P F 1,29 P Initial egg mass (g) Final egg mass (g) < Incubation period (days) Snout vent length (mm) Head length (mm) Tail length (mm) Body mass (g) Maximum strength (g) < Mean strength (g) Standard deviation of strength (g) The data comprise F- and P-values from a two-factor analysis of variance on traits of eggs and hatchlings. Mean values per treatment are shown in Figure 4. Table 2. Effects of incubation conditions (mean nest temperature and substrate water content) on phenotypic traits of hatchlings of keelback snakes, after correcting for hatchling snout vent length by including this trait as a covariate Mean temperature (warm vs. cool) Moisture content ( vs. ) Interaction Variable F 1,29 P F 1,29 P F 1,29 P Head length (mm) Tail length (mm) Body mass (g) Maximum strength (g) Mean strength (g) Standard deviation of strength (g) The data comprise F- and P-values from a two-factor analysis of variance on traits of hatchlings. became significant (Table 4). The overall MANCOVA showed that, even after the effects of hatchling SVL were removed from the analysis, hatchling morphology and strength were significantly influenced by the diel thermal variance during incubation (F 1,29 = 5.2, P <.2) as well as by the moisture content of the substrate (F 1,29 = 4.5, P <.4). As in previous analyses, no significant interaction between these factors was apparent (F 1,29 = 1.2, P =.37). Thus, incubationinduced modifications to hatchling phenotypes were not simply the result of overall body-size differences among hatchlings from different treatments. DISCUSSION The results of the present study support and extend our previous findings on developmental plasticity in these small tropical snakes. As in earlier studies (Brown & Shine, 24, 25; Shine & Brown, 22), strong effects of substrate moisture content on a range of phenotypic traits were found, and such effects were apparent even after the incubation-induced shift in mean body size of hatchlings was factored out of the analysis. Importantly, the present study used smaller differences in diel thermal variance (8.4 vs. 6. C) than those employed in our earlier experiments (7.8 vs. 1.3 C; Webb et al., 21). Mean temperatures were similar in the two studies (25.4 and 25.9 C in the present study; 25.6 C for both treatments in Webb et al., 21). The present study shows that a 2 C difference in mean incubation temperatures influenced offspring size but had no significant impact on other traits. The production of larger offspring from cooler moister nests, combined with a significant survival

8 166 G. P. BROWN and R. SHINE 16 A.3 B low Snout-vent length (cm) Body condition high C.6 D Relative head length Relative strength Incubation moisture Incubation moisture Figure 5. Mean values and associated standard errors for phenotypic traits of hatchling keelback snakes incubated under different conditions. The traits are are snout vent length (A), body condition (B), relative head length (C), and relative strength (D). Body condition and relative traits are based on residual scores from linear regressions of the trait in question (ln body mass, head length, or mean strength) against ln snout vent length. The eggs were incubated at different diel thermal variances, (diel range low 6. vs. high 8.4 C) and in vermiculite of different substrate moisture contents ( 5% vs. 1% water by weight). Vertical bars indicate standard errors. benefit to larger hatchlings (Brown & Shine, 24), fits well with a trend for female keelbacks to select moist nest-sites offering cooler conditions than are available over most of the study site (Fig. 1). Lastly, our extensive series of factorial analyses detected only a single marginally significant interaction between moisture and temperature (Tables 1, 2, 3, 4), indicating that the phenotypic effects of these two factors are additive rather than synergistic (either negatively or positively). Our MANOVA and MANCOVA analyses all suggested that soil moisture had a greater effect than temperature (either mean or diel variance) on hatchling phenotypes. However, this comparison is sensitive to the details of the experimental treatments; for example, reducing the hydric discrepancy between our moisture treatments, or increasing that between the temperature treatments, might well reverse this pattern. Identifying which is more important will depend upon the range of conditions that a nesting snake encounters in the field, and documenting such a set of choices remains a formidable logistical challenge. The phenotypic modifications induced by the incubation treatments in the present study spanned a wide range of traits, at least one of which (offspring

9 INCUBATION EFFECTS IN A SNAKE 167 Table 3. Effects of incubation conditions (diel variance in nest temperature, and mean substrate water content) on phenotypic traits of eggs and hatchlings of keelback snakes Thermal variance (high vs. low) Moisture content ( vs. ) Interaction Variable F 1,29 P F 1,29 P F 1,29 P Initial egg mass (g) Final egg mass (g) < Incubation period (days) Snout vent length (mm) Head length (mm) Tail length (mm) Body mass (g) Maximum strength (g) Mean strength (g) Standard deviation of strength (g) The data comprise F- and P-values from a two-factor analysis of variance on traits of eggs and hatchlings. Mean values per treatment are shown in Figure 5. Table 4. Effects of incubation conditions (diel variance in nest temperature, and mean substrate water content) on phenotypic traits of hatchlings of keelback snakes, after correcting for hatchling snout vent length by including this trait as a covariate Thermal variance (high vs. low) Moisture content ( vs. ) Interaction Variable F 1,29 P F 1,29 P F 1,29 P Tail length (mm) Head length (mm) Body mass (g) Maximum strength (g) Mean strength (g) Standard deviation of strength (g) The data comprise F- and P-values from a two-factor analysis of variance on traits of hatchlings. body length) is under significant directional selection in this population (Brown & Shine, 24). The ontogenetic persistence and/or microevolutionary significance of incubation-induced shifts in traits such as relative head size and tail length warrant further study. It is easy to imagine circumstances in which such traits might influence an individual s fitness, but the mark recapture data available do not reveal any linkages between these traits and fitness (G. P. Brown & R. Shine, unpubl. data). The conclusion that mean incubation temperature, as well as diel thermal variance, influences the phenotypic traits of offspring mirrors the results of studies on many other squamate reptiles, including the sympatric water python Liasis fuscus (Shine et al., 1996; Deeming, 24). However, the low thermal optimum for embryogenesis (note that cooler incubation gave rise to larger hatchlings; Fig. 4) is surprising for a tropical reptile. One possible explanation for this pattern involves the covariation between hydric and thermal attributes of potential nest-sites (Fig. 3) (i.e. moist nests will often be cooler and more thermostable than nests). At Fogg Dam, such correlations are generated by exposure to sunlight (exposed nests are hotter, more variable thermally and drier) and soil depth (at any given site, deeper nests are moister and less variable thermally; Fig. 3). Offspring fitness in keelbacks is maximized by oviposition in a moist nestsite, and reproducing females select such sites when given a choice (Brown & Shine, 24). Such a preference automatically will tend to place the eggs in cool, thermally stable locations. In turn, this should favour the evolutionary adjustment of norms of reaction to optimize embryogenesis under cool and thermostable

10 168 G. P. BROWN and R. SHINE as well as moist conditions. If one abiotic factor (in this case, soil moisture) that strongly affects offspring fitness typically co-occurs with another factor (in this case, thermal regimes) in natural nests, then embryonic adaptations to incubation under such thermal (as well as hydric) conditions are expected. In keeping with this hypothesis, the data from the present study on keelbacks suggest that cooler nests increase offspring size, as do moister nests (Figs 4, 5). Thus, if selection favours oviposition in moister nest-sites (as occurs in the present study population), evolution would be expected to favour an optimal incubation temperature corresponding to that encountered in hydrically optimum nests. In the case of keelback snakes, this linkage may have led to the evolution of a surprisingly low thermal optimum for incubation despite the wide availability of much hotter nest-sites. ACKNOWLEDGEMENTS We thank C. Shilton, E. Cox, and the staff of Beatrice Hill farm for their encouragement and assistance, and the Australian Research Council for funding. REFERENCES Ballinger RE Life-history variations. In: Huey RB, Pianka ER, Schoener TW, eds. Lizard ecology: studies of a model organism. Cambridge, MA: Harvard University Press, Birchard GF. 24. Effects of incubation temperature. In: Deeming DC, ed. Reptilian incubation. Environment, evolution and behaviour. Nottingham: Nottingham University Press, Brown GP, Shine R. 22. Reproductive ecology of a tropical natricine snake, Tropidonophis mairii (Colubridae). Journal of Zoology (London) 258: Brown GP, Shine R. 24. Maternal nest-site choice and offspring fitness in a tropical snake (Tropidonophis mairii, Colubridae). Ecology 85: Brown GP, Shine R. 25. Do changing moisture levels during incubation influence phenotypic traits of hatchling snakes (Tropidonophis mairii)? Physiological and Biochemical Zoology 78: Cogger HG Reptiles and amphibians of Australia, 4th edn. Sydney: Reed Books. Deeming DC. 24. Post-hatching phenotypic effects of incubation on reptiles. In: Deeming DC, ed. Reptilian incubation. Environment, evolution and behaviour. Nottingham: Nottingham University Press, Flatt T, Shine R, Borges-Landaez PA, Downes SJ. 21. Phenotypic variation in an oviparous montane lizard (Bassiana duperreyi): the effects of thermal and hydric incubation environments. Biological Journal of the Linnean Society 74: Ji X, Brana F The influence of thermal and hydric environments on embryonic use of energy and nutrients, and hatchling traits, in the wall lizards (Podarcis muralis). Comparative Biochemistry and Physiology A 124: Madsen T, Shine R Seasonal migration of predators and prey: pythons and rats in tropical Australia. Ecology 77: Muth A Physiological ecology of desert iguana (Dipsosaurus dorsalis) eggs: temperature and water relations. Ecology 61: Packard GC, Miller K, Packard MJ Environmentally induced variation in body size of turtles hatching in natural nests. Oecologia 93: Rhen T, Lang JW Phenotypic plasticity for growth in the common snapping turtle: effects of incubation temperature, clutch, and their interaction. American Naturalist 146: Scheiner SM, Callahan HS Measuring natural selection of phenotypic plasticity. Evolution 53: Seigel RA, Ford NB Phenotypic plasticity in the reproductive characteristics of an oviparous snake, Elaphe guttata: implications for life history studies. Herpetologica 47: Shine R Strangers in a strange land ecology of the Australian colubrid snakes. Copeia 1991: Shine R. 24. Adaptive consequences of developmental plasticity. In: Deeming DC, ed. Reptilian incubation. Environment, evolution and behaviour. Nottingham: Nottingham University Press, Shine R, Brown GP. 22. Effects of seasonally varying hydric conditions on hatchling phenotypes of keelback snakes (Tropidonophis mairii, Colubridae) from the Australian - tropics. Biological Journal of the Linnean Society 76: Shine R, Madsen T, Elphick M, Harlow P The influence of nest temperatures and maternal thermogenesis on hatchling phenotypes of water pythons. Ecology 78: Sinervo B, Svensson E Mechanistic and selective causes of life history trade-offs and plasticity. Oikos 83: Sultan S Evolutionary implications of phenotypic plasticity in plants. Evolutionary Biology 21: Via S, Gomulkiewicz R, De Jong G, Scheiner SM, Schlichting CD, Van Tienderen PH Adaptive phenotypic plasticity: consensus and controversy. Trends in Ecology and Evolution 1: Via S, Lande R Genotype environment interaction and the evolution of phenotypic plasticity. Evolution 39: Webb JK, Brown GP, Shine R. 21. Body size, locomotor speed and antipredator behaviour in a tropical snake (Tropidonophis mairii, Colubridae): the influence of incubation environments and genetic factors. Functional Ecology 15: