and hydration of hatchling Painted Turtles, Chrysemys picta

Functional Ecology 21 Environmentally induced variation in size, energy reserves Blackwell Science, Ltd and hydration of hatchling Painted Turtles, Chrysemys picta G. C. PACKARD and M. J. PACKARD Colorado State University, Department of Biology, Fort Collins, CO 8523-1878, USA Summary 1. The contents of newly constructed nests of Painted Turtles, Chrysemys picta (Schneider 1783), were manipulated by reciprocal transplant so that each of several nests received a complement of eggs from each of several females. The eggs were recovered from nests after 8 weeks and allowed to complete their incubation under standard conditions in the laboratory. The design of the experiment enabled us to distinguish between environmental and maternal effects on attributes of hatchlings. 2. Several measures of body size and energy reserve varied among turtles hatching from eggs that incubated in different nests, and certain of these measures varied also among turtles hatching from eggs that incubated in different layers within nests. The effects of nest and layer were substantial. For example, fat-free carcasses of hatchlings from one nest weighed 17% more than those of neonates from a second nest, but fatfree yolks from the former weighed only 53% as much as yolks from the latter. 3. Stepwise linear regression indicated that the size of hatchlings and the hydration and fat content of their carcasses were positively correlated with the net change in mass of eggs (which is an index to net water-exchange) while they incubated in the field. In contrast, both the fat and fat-free components of unused yolk were negatively correlated with change in mass of eggs. Although the statistical procedure is only correlative, the findings accord well with results of laboratory studies documenting a relationship between uptake of water by eggs, metabolism and growth by embryos, and size and condition of hatchlings. 4. Variation among hatchlings representative of different nests accounted for 24% of the statistical variance in mass of dry, fat-free carcasses; 29% of the variance in mass of dry, fat-free yolks; 19% of the variance in mass of storage fat in yolks; and 11% of the variance in mass of storage fat in carcasses. Additional variation was detected between the upper and lower layers of nests. Such environmentally induced variation probably affects survival of neonatal animals in the field. Key-words: Embryo, incubation, plasticity, turtle Functional Ecology (21) Ecological Society Introduction A recent investigation from our laboratory focused on the role of the nest environment in eliciting variation in body size and condition in neonatal Painted Turtles, Chrysemys picta (Schneider 1783) (see Packard & Packard 2). The study entailed transplanting eggs reciprocally among several natural nests so as to distinguish clearly between environmental effects and those related to clutch of origin (which is an amalgam of genetic effects and maternal effects). The eggs were Author to whom correspondence should be addressed. E-mail: packard@lamar.colostate.edu allowed to incubate in the field for approximately 8 weeks, and then were recovered from the nests and brought into the laboratory to complete incubation under optimal conditions of temperature and water potential. The study revealed that an important fraction of the variation in live mass of hatchlings, dry mass and water content of their carcasses, and dry mass of their unused (residual) yolks resulted from variation in the physical environment within and among nests during the time that eggs were incubating in the field. Eggs in some nests absorbed water (net) and increased in mass during that time, whereas eggs in other nests lost water to their surroundings and declined in mass. Embryos having access to relatively large amounts of 481

482 G. C. Packard & M. J. Packard water seemingly sustained higher rates of metabolism and growth than embryos having access to relatively small amounts of liquid, and the differences in metabolism and growth translated into the observed differences in size and condition of hatchlings. However, the preceding investigation yielded an incomplete picture of environmentally induced variation in size and condition of hatchlings, because it was not possible to characterize the amounts of lipid and protein in residual yolks or the amount of storage fat in carcasses. Neutral fats were therefore extracted from residual yolks and carcasses of turtles from the aforementioned study to obtain new insights concerning the influence of environmental factors on the size and distribution of energy reserves in neonates. The new findings are reported here. Materials and methods ANIMALS AND MANIPULATIONS The field component of this experiment was performed at the Valentine National Wildlife Refuge, Cherry County, Nebraska, USA, during the summer of 1999 (see Packard & Packard 2). The contents of newly constructed nests of Painted Turtles were manipulated by a reciprocal transplant procedure, so that each of four nests received two eggs from each of the same five females. One weighed egg from each female was placed in a bottom layer of each nest, and another was assigned to a top layer. This basic procedure was replicated on three consecutive days, so that the investigation was based at the outset on a sample of 12 eggs, 15 clutches and 12 nests. The thermistor probe of a StowAway datalogger (Onset Computer Corp., Pocasset, MA, USA) also was placed at the bottom of each nest to record temperature at intervals of 3 min. The eggs were allowed to incubate in the field for 54 56 days (depending on the replicate), at the end of which the nests were re-opened and the eggs were recovered and re-weighed. The settling of soil into egg chambers caused the eggs in several nests to be shifted in position from where they were at the outset of study. Accordingly, the first five eggs that were encountered as we dug vertically downwards into each nest were taken to represent the top layer, and the remaining eggs were taken to represent the bottom layer. Three of the nests (one from each replicate) were empty, owing presumably to predation on the eggs by a snake. Eggs from the remaining nine nests were transported to the laboratory and incubated to hatching on moistened vermiculite (water potential of 1 kpa) at 26 C. A total of 74 healthy turtles ultimately hatched from the eggs and served as the basis for this study. Turtles were killed by freezing on the day after hatching, and the residual yolk was dissected from the abdomen of each animal. The yolk and carcass were dried individually at 5 C (Kerr, Ankney & Millar 1982), weighed on an electronic balance, and then stored in a freezer until neutral fats could be extracted with petroleum ether. The extractions were accomplished with a Goldfisch apparatus operating for 5 h ( Dobush, Ankney & Krementz 1985). The extracted yolks and carcasses then were re-dried and re-weighed, and the mass of extracted fat in each was determined by subtraction. STATISTICAL ANALYSES The data for initial mass of eggs, for dry mass of fatfree carcasses and yolks, for fat content of the carcasses and yolks separately, and for total fat available to each animal in yolk plus carcass were first examined with ANCOVAs in a split-plot design. The analyses were performed using the Mixed Procedure in SAS version 8 1 (SAS Institute 1996), and treated nest and clutch as fixed effects nested within day, which was treated as a random effect. Layer was a fixed effect across days, and initial mass of eggs was a potential covariate (except in the analysis for egg mass). A similar ANCOVA was performed on data for body water, but mass of the fat-free carcass was the covariate to remove effects of variation in body size within treatment groups. The only interaction term considered in these analyses was that between layer and nest, because this is the only interaction that can be interpreted meaningfully in an ecological context and because other interactions were found in our earlier investigation not to be important sources of variation ( Packard & Packard 2). In instances where the covariate proved not to be a significant source of variation, we reverted to ANOVA for the final analysis. These several ANCOVAs (or ANOVAs) enabled us to generate least squares means so that we could assess the magnitude of the variation among the specific set of clutches and nests used in this study. The data were studied next by stepwise (forward) linear regression (Proc Reg in SAS). The goal here was to gain insight concerning causes for the variation detected among nests. Mean temperatures in nests (see Packard & Packard 2) and net change in mass of eggs over 8 weeks of incubation in the field were potential independent variables in these analyses. Clutch of origin also was treated as a potential independent variable by constructing a matrix of dummy variables and by treating the matrix as a set in the stepping procedure (Cohen 1991). Mass of the fat-free carcass was a fourth independent variable in the analysis of body water. Finally, another set of ANCOVAs (or ANOVAs) was performed on data for carcass and yolk of hatchlings. The design for these final analyses was virtually identical to that used in the earlier ones, except that nest and clutch were treated this time as random factors. Consequently, the data were modelled differently by the statistical algorithm from the way they were modelled in the initial series of ANCOVAs (SAS Institute 1996), resulting in some small differences in the outcome. This

483 Energy reserves of hatchling turtles treatment enabled us to estimate variances and thereby predict the relative importance of these factors as sources of variation among hatchlings emerging from nests in the field (Beck 1997). This is also the analysis that was used to assess the importance of layer (a fixed factor) and to generate least squares means for eggs incubating at the top and bottom of nests. Day was treated again as a random factor. Levels of significance for components of variance were estimated by invoking the test option for random factors in Proc GLM of SAS (199). Results MAGNITUDE OF VARIATION AMONG NESTS AND CLUTCHES Analysis of variance revealed that clutch was the only source of variation in data for initial mass of eggs (Table 1). This outcome indicates that representative samples of eggs went into nests prepared on each of the three days in June (Table 2). However, the variation among clutches in size of eggs was substantial, with clutches that were sampled on the same day differing in average mass by more than 1 7 g (Table 3). It should be noted here that comparisons need to be limited to nests and clutches prepared for study on the same day, because samples taken on different days are not necessarily comparable. Both nest and clutch influenced the mass of dry, fat-free carcasses of hatchlings, the amount of water in their carcasses, the mass of dry, fat-free yolks removed from the turtles, and the mass of neutral fats in those yolks ( Table 1). The differences detected among clutches generally were greater than those detected among nests sampled the same day (Tables 2, 3), but effects of nest were nonetheless substantial. For example, fat-free carcasses of hatchlings from nest 53 weighed 17% more than those of neonates from nest 5, whereas fat-free yolks from the former weighed only 53% as much as yolks from the latter (Fig. 1). The reserve of fats in yolk was only 42% as large in animals from nest 53 as it was in turtles from nest 5 (Fig. 1). Table 1. F-ratios and levels of significance for the fixed effects in mixed-model analyses of variance (or covariance) performed on data for eggs and hatchlings of Painted Turtles. Clutch and nest were treated as fixed factors for purposes of these analyses. The potential covariate was initial mass of eggs except for the analysis of data for carcass water, where the covariate was mass of the fat-free carcass. Denominator df = 44 for ANOVAs, 43 for the ANCOVAs Source of variation df Initial egg mass Fat-free carcass Carcass water Fat-free yolk Yolk fat Carcass fat Total fat Clutch (day) 12 62 84 5 13 Nest (day) 6 37 5 83 ( 896) Layer 1 1 95 2 11 ( 17) ( 154) Nest layer (day) 8 61 1 22 ( 767) ( 312) Covariate 1 NA 7 27 ( 1) 4 68 3 2 ( 11) 6 54 ( 14) 1 11 ( 375) 47 39 5 3 7 49 7 27 29 69 7 73 4 79 1 85 2 22 ( 112) ( 59) 4 11 3 28 16 2 42 ( 49) ( 77) ( 694) ( 127) 78 53 39 27 ( 625) ( 824) ( 923) ( 973) NS NS NS NS Table 2. Least squares means for initial mass of eggs from different nests (g), for dry mass and water content of fat-free carcasses (mg), for dry mass of fat-free yolks (mg), and for mass of neutral fats from yolks, carcasses, and yolks plus carcasses (mg). Data for mass of the fat-free carcass were adjusted by ANCOVA to remove effects of variation in size of eggs (Table 1), and values for carcass water were adjusted by ANCOVA to remove effects of mass of fat-free carcasses. No adjustment was required for other data Variable Day Nest N Egg mass Fat-free carcass Carcass water Fat-free yolk Yolk fat Carcass fat Total fat 1 35 9 6 481 a 73 a 3555 a 69 a 67 a 175 a 243 a 1 36 6 6 451 a 699 a 3473 a 98 a 77 a 166 a 243 a 1 39 7 6 564 a 743 a 3422 a 61 a 65 a 166 a 231 a 2 5 7 6 829 a 675 a 3542 a 169 a 97 a 175 a 272 a 2 52 1 6 835 a 768 b 3652 a,b 72 b 55 b 191 a,b 246 b 2 53 1 6 78 a 788 b 3711 b 79 b 56 b 24 b 26 a 3 62 9 6 216 a 778 a,b 3555 a 71 a 43 a 182 a 225 a 3 64 1 6 195 a 747 b 3343 b 94 a 51 a 173 a 223 a 3 65 6 6 288 a 82 a 346 a,b 77 a 43 a 175 a 218 a Least squares means for nests prepared on a given day cannot be distinguished statistically (P = 5) if they are joined by the same superscript. See also Table 1 to determine which comparisons are warranted.

484 G. C. Packard & M. J. Packard Table 3. Least squares means for initial mass of eggs from different clutches (g), for dry mass and water content of fat-free carcasses (mg), for dry mass of fat-free yolks (mg), and for mass of neutral fats from yolks, carcasses, and yolks plus carcasses (mg). Data for mass of the fat-free carcass were adjusted by ANCOVA to remove effects of variation in size of eggs (Table 1), and values for carcass water were adjusted by ANCOVA to remove effects of mass of fat-free carcasses. No adjustment was required for other data Variable Day Clutch N Egg mass Fat-free carcass Carcass water Fat-free yolk Yolk fat Carcass fat Total fat 1 33 5 7 572 a 589 a 3664 a 74 a 65 a,b 196 a 261 a 1 34 4 6 28 b 743 b 3429 b,c 6 a 74 a,b 148 b,c 222 b 1 36 5 6 369 c 774 b 3477 b,c 89 a 84 b 176 a,b 26 a 1 37 5 6 784 d 794 b 3574 a,b 11 a 51 a 181 a 232 b 1 39 3 5 741 b 717 a,b 3271 c 57 a 76 a,b 144 c 219 b 2 5 6 8 6 a 721 a,b 391 a 122 a 94 a,c 233 a 326 a 2 51 5 6 734 b 743 a,b 3692 b 54 b 26 b 16 b 188 b 2 52 6 6 275 c 73 a 3579 b,c 8 b 41 b 161 b 22 b 2 53 5 6 644 b,d 768 b 3544 b,d 91 a,b 77 a 213 a 29 c 2 54 5 6 414 c,d 785 b 345 c,d 187 c 16 c 185 c 291 c 3 61 5 6 836 a 783 a 352 a,d 88 a,b 36 a,c 175 a,b 21 a 3 62 6 6 428 b 741 a 3394 b,c,d 12 b 67 b 18 a,b 247 b 3 63 5 5 115 c 781 a 3244 b 59 a 47 b,c 166 b 213 a 3 64 5 6 796 a 792 a 367 a 97 a,b 42 a,c 197 a 239 b 3 66 4 5 99 d 782 a 3498 a,c 57 a 37 a,c 164 b 21 a Least squares means for clutches prepared on a given day cannot be distinguished statistically (P = 5) if they are joined by the same superscript. Table 4. Least squares means for different measures of size and condition for Painted Turtles hatching from eggs that incubated for 8 weeks in different layers of nests in the field. Data for fat-free carcass, carcass fat and total fat were adjusted by ANCOVA to remove effects of variation in initial size of eggs, and those for carcass water were adjusted to remove effects of variation in mass of the fat-free carcass. No adjustment was required for data for fat-free yolk or yolk fat. Clutch and nest were treated as random factors in these analyses Variable Upper layer Lower layer Level of significance Fat-free carcass (mg) 739 753 285 Carcass water (mg) 349 3583 42 Fat-free yolk (mg) 95 8 69 Yolk fat (mg) 64 58 152 Carcass fat (mg) 181 18 824 Total fat (mg) 245 238 86 Fig. 1. Least squares means (± 2 SEM) for masses of body components for Painted Turtles hatching from eggs that incubated for 8 weeks in nests 5 and 53. The neonates from these two nests differed more in most measures of carcass composition than did animals from any other pair of nests prepared for study on the same day. The amount of fat in carcasses did not vary among the specific set of nests used in this investigation, and the measure of total fat varied only slightly (Table 1; Fig. 1). These same variables were affected significantly by clutch, however. INFLUENCE OF LAYER IN NEST Hatchlings from eggs that incubated in the lower layer of cnests were better hydrated than turtles emerging from eggs that incubated in the upper layer (Table 4), and both the fat-free yolk and the total reserve of fat available to hatchlings were smaller in animals representing the bottom layer (Table 4). Variation between layers was not clear-cut for mass of the fat-free carcass or for the amount of fat extracted from yolk or carcass ( Table 4). CAUSES FOR VARIATION Clutch entered first into regression models for mass of fat-free carcasses as well as for the several measures of

485 Energy reserves of hatchling turtles Table 5. F-ratios, levels of significance and model R 2 for independent variables entering step-wise regression models for mass of fat-free carcass and yolk, mass of water in the carcass, mass of fat in yolk and carcass, and total mass of the fat reserve available to neonatal Painted Turtles Independent variable (df ) Step 1 Step 2 Step 3 Fat-free carcass Clutches (14) F = 6 78 R 2 = 617 egg mass (1) F = 3 4 P = 69 R 2 = 45 Mean temperature (1) F = 47 P = 496 R 2 = 7 Carcass water Clutches (14) F = 8 26 R 2 = 662 Fat-free carcass (1) F = 24 78 R 2 = 74 egg mass (1) F = 3 35 P = 71 R 2 = 45 Mean temperature (1) F = 5 26 P = 25 R 2 = 68 Fat-free yolk Clutches (14) F = 3 17 P = 1 R 2 = 429 egg mass (1) F = 32 16 R 2 = 39 Mean temperature (1) F = 37 P = 545 R 2 = 5 Yolk fat Clutches (14) F = 6 34 R 2 = 61 egg mass (1) F = 17 7 R 2 = 192 Mean temperature (1) F = 2 87 P = 95 R 2 = 38 Carcass fat Clutches (14) F = 7 28 R 2 = 634 egg mass (1) F = 1 P = 97 R 2 = 1 Mean temperature (1) F = 55 P = 462 R 2 = 8 Total fat Clutches (14) F = 29 44 R 2 = 875 egg mass (1) F = 8 41 P = 5 R 2 = 15 Mean temperature (1) F = 2 99 P = 88 R 2 = 4 F = 26 51 R 2 = 737 F = 21 P = 648 R 2 = 618 F = 88 29 R 2 = 866 F = 39 98 R 2 = 8 F = 3 35 P = 72 R 2 = 681 F = 42 44 R 2 = 671 F = 2 P = 893 R 2 = 43 F = 23 47 R 2 = 716 F = 15 P = 74 R 2 = 62 F = 9 9 P = 4 R 2 = 683 F = P = 96 R 2 = 634 F = 4 5 P = 49 R 2 = 883 F = 3 P = 585 R 2 = 875 F = 1 P = 926 R 2 = 737 F = 1 25 P = 2 R 2 = 887 F = 5 54 P = 22 R 2 = 878 F = 1 1 P = 32 R 2 = 676 F = 1 15 P = 289 R 2 = 721 F = 11 P = 747 R 2 = 684 F = 68 P = 414 R 2 = 884 neutral fat, and clutch (which was a significant variable anyway) was forced to enter first into the models for carcass water and fat-free yolk ( Table 5). Thus, the first step in the stepping procedure effectively accounted for the influence of clutch on size and energy status of hatchlings. Change in mass of eggs over 8 weeks of incubation in the field (delta) was superior to mean temperature in nests at explaining residual variation in dependent variables other than carcass water, so change in mass entered most of these regression models at the second step (Table 5). The residuals for masses of carcasses and extracted fats were positively correlated with change in mass of eggs, whereas the corresponding values for yolks were negatively correlated with change in mass of eggs (Fig. 2). Residuals for the total reserve of neutral fats available in yolk plus carcass also was correlated negatively with change in mass of eggs during incubation in the field (Fig. 2). Temperature did not explain a significant amount of variation once change in mass of eggs had entered into the models (Table 5). Mass of the fat-free carcass entered at the second step into the regression model for carcass water, so the first two steps in this regression analysis effectively accounted for variation stemming from clutch and from body size within clutch (Table 5). Both change in mass of eggs and nest temperature explained a significant amount of the remaining variation, but more of this residual variation was explained by the former than by the latter (Table 5). Thus, change in mass of eggs entered the model at the third and final step. Eggs that absorbed large quantities of water from their environment and increased in mass produced hatchlings that were relatively well hydrated, whereas eggs that lost water to their surroundings and declined in mass produced hatchlings that were slightly desiccated (Fig. 2). SOURCES OF VARIATION IN THE POPULATION AT LARGE Clutch of origin was the only important source of variation in initial mass of eggs other than the unexplained (= residual) variance (Table 6). The absence of a significant component of variance attributable to nest again indicates that each nest received a representative sample of eggs (i.e. no bias was introduced by our protocol). These analyses also revealed that neither the day on which a set of nests was prepared nor the interaction between nest and layer contributed significantly to variance in any of the dependent variables (Table 6). However, both nest and clutch contributed significantly to variance in all the measures of size and condition, albeit the influence of nest on carcass water was not compelling ( Table 6). By setting the non-significant sources of variation to zero, it can be seen that nest contributed 24% to statistical variance in data for fatfree carcasses, 29% to variance in values for fat-free yolk, 19% to variance in values for fat in yolk, and 11% to

486 G. C. Packard & M. J. Packard from carcasses and yolks of hatchling Painted Turtles in the current study were storage lipids that animals could have drawn upon to support their metabolism (Hadley 1985), and our measures of total fat are indices to the size of the lipid reserve available to the neonatal turtles. Structural lipids (phospholipids) are not removed by petroleum ether (Dobush et al. 1985), so these materials constituted an undetermined fraction of the mass of extracted yolks in this study. However, phospholipids typically constitute only a minor fraction of lipids in yolks taken from fresh eggs or hatchlings (Rowe et al. 1995). The bulk of the fat-free yolk from turtles in this study therefore presumably was protein that neonates could have used in the synthesis of new tissue or as a substrate for energy transformations ( Wilhoft 1986; Janzen et al. 199). Structural lipids comprised a small, undetermined fraction of the mass of extracted carcasses, but the mix of constituents in carcasses also included mineral (in bone and shell), protein and carbohydrate. Indeed, little can be said about the composition of carcasses, except that they did not contain storage lipids. Accordingly, the mass of the dry, lipid-free carcasses probably should be taken only to reflect on overall size of the animals. ENVIRONMENTALLY INDUCED VARIATION Fig. 2. Plots of residuals for dry mass of fat-free carcasses and yolks taken from hatchling Painted Turtles, for water content of carcasses, for mass of neutral fats in the unused yolks and carcasses, and for total mass of lipid in yolks plus carcasses in relation to net change in mass of eggs over 8 weeks of incubation in nests in the field. The residuals are from regression models that used clutch as a dummy, or location, variable to remove effects of clutch from each of the dependent variables. The regression model for carcass water also included mass of the fat-free carcass as an independent variable, so the residuals plotted here reflect on variation in the level of hydration of the tissues. The simple correlation coefficient is given in each panel. variance in data for fat in carcasses. Corresponding contributions by clutch were 38%, 33%, 48%, and 32%, respectively. Nest accounted for 13% of the variance in data for water in carcasses of hatchlings and 4% of the variance in values for total fat ( Table 6). The contributions of clutch were correspondingly larger 43% and 79%, respectively. Discussion BODY COMPONENTS Petroleum ether is a non-polar solvent that dissolves neutral fats (Dobush et al. 1985), most of which are triacylglycerols (Rowe et al. 1995). Thus, the lipids extracted Effects attributable to nest are effects attributable to the environment ( Table 1). Different temperatures and moisture regimes in different nests presumably elicited different patterns of physiological response from developing embryos (Morris et al. 1983; Gettinger, Paukstis & Gutzke 1984; Packard & Packard 1986; Janzen et al. 199; Miller & Packard 1992), so that embryos from some nests grew to larger size before hatching than did embryos from other nests (Table 2; Fig. 1). In the process of growing to relatively large size, these embryos mobilized more of the nutrient reserves in their yolks. Consequently, a smaller reserve of fats and protein remained in the yolk to sustain hatchlings in the neonatal period (Table 2; Fig. 1). Components of variance computed for random factors in an ANOVA yield estimates for the importance of these factors as sources of variation in the population at large ( Beck 1997). We therefore anticipate that approximately one-quarter of the variation in mass of the fatfree carcass of turtles hatching in different nests in the field results from developmental plasticity ( Table 6). Additionally, 13% of the variation in hydration of carcasses is induced by variable nest environments; 29% of the variation in mass of the fat-free yolk; 19% of the variation in fat in unused yolk; and 11% of the variation in carcass fat (Table 6). The contributions of clutch to total variance were greater than those of nest (Table 6), owing in large measure to the very substantial variation in size of eggs produced by different females (Table 3). Indeed, the

487 Energy reserves of hatchling turtles Table 6. Components of variance for random factors in analyses of data for hatchling Painted Turtles that completed 8 weeks of incubation in natural nests Source of variance Variable Day Nest (day) Clutch (day) Nest layer (day) Residual Initial mass for eggs F 2,11 8 = 1 4 P = 384 Fat-free carcass F 2,15 = 73 P = 496 Carcass water 281 F 2,13 6 = 2 3 P = 17 Fat-free yolk F 2,11 6 = 5 P = 621 Yolk fat 18 F 2,15 1 = 1 19 P = 331 Carcass fat F 2,17 2 = 92 P = 417 Total fat F 2,14 3 = 44 P = 652 F 6,7 1 = 62 P = 712 1253 F 6,7 5 = 4 75 P = 27 4658 F 6,8 4 = 2 89 P = 8 775 F 6,7 3 = 1 5 P = 3 171 F 6,7 = 9 23 P = 5 58 F 6,6 5 = 5 5 P = 24 53 F 6,5 8 = 9 68 P = 8 4839 F 12,44 = 62 84 1987 F 12,43 = 5 13 15 27 F 12,43 = 4 68 897 F 12,44 = 5 3 443 F 12,44 = 7 49 178 F 12,43 = 3 34 P = 2 186 F 12,43 = 22 31 F 8,44 = 61 P = 767 119 F 8,43 = 1 22 P = 312 1159 F 8,43 = 1 11 P = 375 F 8,44 = 78 P = 625 F 8,44 = 53 P = 824 F 8,43 = 41 P = 91 F 8,43 = 27 P = 972 32 195 15 771 112 31 316 228 eggs from clutches 33, 5 and 61 were, on the average, 1 7 1 8 g (or 3%) heavier than eggs from clutches 39, 52 and 63, respectively (Table 3). Large eggs contain large yolks (i.e. large energy reserves in the form of protein and neutral fats) with large reserves of water (Finkler & Claussen 1997), and both of these contribute to the production of large hatchlings ( Finkler 1997). To the extent that variance due to clutch resulted from variation in size of eggs produced by different females, this variation was arguably induced also by the environment ( Bernardo 1996) and resulted from differences among females in their size and level of nutrition. Nevertheless, environmentally induced variation of this sort is qualitatively different from that induced by the physical environment inside nests. IMPORTANCE OF LAYER Effects attributable to layer also are effects induced by the environment (Table 4). Temperatures inside a nest usually are lower, on the average, and less variable near the bottom than at the top (Thompson 1988; Ratterman & Ackerman 1989; Georges 1992), and water potential typically increases with depth in the soil (Ackerman 1991). Results of laboratory experiments on the flexibleshelled eggs of Painted Turtles and Snapping Turtles, Chelydra serpentina (Linnaeus 1758) indicate that embryos in cool, wet sites that is, at greater depth in nests should mobilize more of the fats and proteins from their egg yolk and grow to a larger size before hatching than embryos in warm, dry locations (Gutzke et al. 1987; Packard et al. 1987, 1988). This prediction is born out in some field studies (Packard, Miller & Packard 1993; Packard et al. 1999), although results of the current investigation appear on first examination to be less than compelling (Table 4). However, nests of Painted Turtles do not have a pronounced vertical dimension, so eggs in the bottom layer are not much further from the surface than eggs in the upper layer. Thus, the physical conditions to which eggs are exposed in the bottom layer of any given nest are unlikely to differ appreciably from those encountered by eggs in the upper layer, and the impact of the differences is likely to be subtle. It therefore is instructive to re-examine data in Table 4 with these considerations in mind (Warren 1986). Excepting values for carcass fat, all the least squares means in Table 4 both those that differ at accepted levels of significance and those that do not are in the directions expected for turtles hatching from eggs incubating in warm, dry conditions at the top vs cool, moist conditions at the bottom (Gutzke et al. 1987; Packard et al. 1988). Fat-free carcasses were heavier for animals representing the bottom layer, and these turtles were better hydrated than those hatching from eggs nearer the surface ( Table 4). Moreover, turtles from the bottom layer had smaller reserves of neutral fat and protein in their unused yolk, presumably because more of these materials had been mobilized during development to support growth in size. One consequence was a smaller reserve of fat to support these generally larger hatchlings during the neonatal period. Thus, our findings actually lend modest support to the notion that metabolism and growth are elevated among chelonian embryos developing in lower layers, even in nests as small as those of Painted Turtles.

488 G. C. Packard & M. J. Packard CAUSE OF VARIATION Stepwise regression yielded strong, correlative evidence that environmentally induced variation in attributes of hatchling Painted Turtles is linked to water exchanges by their porous, flexible-shelled eggs ( Table 5). Once variation resulting from differences in parentage (and from body size within clutch in the case of data for carcass water) had been removed from the data, the remaining (residual) variation was highly correlated with change in mass of eggs during 8 weeks of incubation in the field (Fig. 2). Change in mass of eggs is a useful (albeit crude) index to the pattern of net water exchange between an egg and its environment, so the pool of available water was augmented in eggs that increased in mass and depleted in eggs that decreased in mass. Embryos having access to relatively large reserves of water presumably had higher rates of metabolism and growth than did embryos with more limited supplies of water ( Morris et al. 1983; Gettinger et al. 1984; Packard & Packard 1986; Janzen et al. 199; Miller & Packard 1992). This caused embryos with a relative abundance of water to consume more of the neutral fat and protein in their yolk and attain larger size before hatching (Fig. 2). Embryos having access to large reserves of water also were better hydrated than embryos with more restricted supplies of water (Fig. 2), and this too is a result that conforms with expectations based on laboratory research ( Morris et al. 1983; Packard et al. 1988). Mean temperature in nests was not particularly useful in explaining variation in size and condition of hatchlings (Table 5), despite the fact that the same measures typically are correlated with temperature in laboratory studies (Gutzke et al. 1987; Packard et al. 1988). As noted earlier ( Packard & Packard 2), this outcome probably reflects on the fact that mean temperature at the bottom of a nest is unlikely to be representative of temperatures experienced by eggs throughout the nest cavity (Packard 1999). SIGNIFICANCE OF DEVELOPMENTAL PLASTICITY Developmental plasticity in Painted Turtles and other chelonians is arguably an adaptive process that matches the demand for water with the availability of water. When water in an oviposited egg is augmented substantially by water that is taken up from the environment, the embryo sustains high rates of metabolism and growth (Morris et al. 1983; Gettinger et al. 1984; Packard & Packard 1986; Janzen et al. 199; Miller & Packard 1992), attains a large size before hatching (Gutzke et al. 1987; Packard et al. 1987) and invests a substantial amount of water in tissues of its body ( Morris et al. 1983; Packard et al. 1988). Conversely, when water is lost from an egg, thereby depleting the reservoir that is available to support the developing embryo, the animal sustains lower rates of metabolism and growth, is smaller at hatching, and invests smaller amounts of water in its tissues. If the process of development were highly canalized and insensitive to the availability of water, an embryo developing in a dry environment might sustain a high rate of metabolism and growth, but it also might exhaust its reservoir of water before attaining the minimum size and stage of development necessary for hatching. The probable outcome consequently would be the death of the embryo. Indeed, even with a down-regulation of metabolism by embryos developing in dry environments, these embryos are more likely than those in wetter settings to die before hatching ( Packard et al. 1989, 1991; Cagle et al. 1993), and deaths of the former generally occur within days, hours or even minutes of hatching. The chorion and amnion of dead turtles usually are dry and extremely adherent to bodies of the animals, which appear subjectively to be desiccated (G. C. Packard & M. J. Packard, unpublished observations). In addition, the different environments occurring within and among nests account for an important fraction of the variation in size of turtles at hatching, in the level of hydration of their tissues, and in the size of the energy reserve (both fats and proteins) that is available to support metabolism and growth during the neonatal period. The potential impact of such variation on survival by hatchlings is currently debated (Congdon et al. 1999; Packard 1999), but recent evidence points to higher survival by relatively large hatchlings than by relatively small ones (Haskell et al. 1996; Janzen et al. 2a, 2b; Tucker 2). Large size may benefit hatchlings by reducing their susceptibility to predation by birds (Janzen et al. 2b). Moreover, the relatively well-hydrated turtles emerging from nests where eggs absorbed large amounts of water may be better able than the slightly dehydrated animals emerging from other nests to withstand stresses associated with movement overland from nest to water (Finkler et al. 2). Taken together, these several considerations support the notion that a wetter environment is better than a drier one for incubating eggs of Painted Turtles (Packard 1999). Acknowledgements This investigation was performed in accord with terms of Special Use Permit VLT 99 1 from the US Fish and Wildlife Service and Scientific Collecting Permit 99 74 from the Nebraska Game and Parks Commission. Our protocol was approved by the Animal Care and Use Committee at Colorado State University (protocol 98 125 A 2). We thank R. R. Huber, M. Lindvall, and L. L. McDaniel for permission to work at the Valentine National Wildlife Refuge, and P. M. Cowley, M. N. Ellsaesser, D. N. Jacobson, C. F. Offermann and H. Whitesides for assistance in the field. The project was supported, in part, by the NSF (IBN 9612562).

489 Energy reserves of hatchling turtles References Ackerman, R.A. (1991) Physical factors affecting the water exchange of buried reptile eggs. Egg Incubation: its Effects on Embryonic Development in Birds and Reptiles (eds D. C. Deeming & M. W. J. Ferguson), pp. 193 211. Cambridge University Press, Cambridge. Beck, M.W. (1997) Inference and generality in ecology: current problems and an experimental solution. Oikos 78, 265 273. Bernardo, J. (1996) Maternal effects in animal ecology. American Zoologist 36, 83 15. Cagle, K.D., Packard, G.C., Miller, K. & Packard, M.J. (1993) Effects of the microclimate in natural nests on development of embryonic Painted Turtles, Chrysemys picta. Functional Ecology 7, 653 66. Cohen, A. (1991) Dummy variables in stepwise regression. American Statistician 45, 226 228. Congdon, J.D., Nagle, R.D., Dunham, A.E., Beck, C.W., Kinney, O.M. & Yeomans, S.R. (1999) The relationship of body size to survivorship of hatchling snapping turtles (Chelydra serpentina): an evaluation of the bigger is better hypothesis. Oecologia 121, 224 235. Dobush, G.R., Ankney, C.D. & Krementz, D.G. (1985) The effect of apparatus, extraction time, and solvent type on lipid extractions of snow geese. Canadian Journal of Zoology 63, 1917 192. Finkler, M.S. (1997) Impact of egg content on post-hatching size, body composition, and performance in the common snapping turtle (Chelydra serpentina). Chelonian Conservation and Biology 2, 452 455. Finkler, M.S. & Claussen, D.L. (1997) Within and among clutch variation in the composition of Chelydra serpentina eggs with initial egg mass. Journal of Herpetology 31, 62 624. Finkler, M.S., Knickerbocker, D.L. & Claussen, D.L. (2) Influence of hydric conditions during incubation and population on overland movement of neonatal snapping turtles. Journal of Herpetology 34, 452 455. Georges, A. (1992) Thermal characteristics and sex determination in field nests of the pig-nosed turtle, Carettochelys insculpta (Chelonia: Carettochelydidae), from northern Australia. Australian Journal of Zoology 4, 511 521. Gettinger, R.D., Paukstis, G.L. & Gutzke, W.H.N. (1984) Influence of hydric environment on oxygen consumption by embryonic turtles Chelydra serpentina and Trionyx spiniferus. Physiological Zoology 57, 468 473. Gutzke, W.H.N., Packard, G.C., Packard, M.J. & Boardman, T.J. (1987) Influence of the hydric and thermal environments on eggs and hatchlings of painted turtles (Chrysemys picta). Herpetologica 43, 393 44. Hadley, N.F. (1985) The Adaptive Role of Lipids in Biological Systems. Wiley, New York. Haskell, A., Graham, T.E., Griffin, C.R. & Hestbeck, J.B. (1996) Size related survival of headstarted redbelly turtles (Pseudemys rubriventris) in Massachusetts. Journal of Herpetology 3, 524 527. Janzen, F.J., Packard, G.C., Packard, M.J., Boardman, T.J. & zumbrunnen, J.R. (199) Mobilization of lipid and protein by embryonic snapping turtles in wet and dry environments. Journal of Experimental Zoology 255, 155 162. Janzen, F.J., Tucker, J.K. & Paukstis, G.L. (2a) Experimental analysis of an early life-history stage: selection on size of hatchling turtles. Ecology 81, 229 234. Janzen, F.J., Tucker, J.K. & Paukstis, G.L. (2b) Experimental analysis of an early life-history stage: avian predation selects for larger body size of hatchling turtles. Journal of Evolutionary Biology 13, 947 954. Kerr, D.C., Ankney, C.D. & Millar, J.S. (1982) The effect of drying temperature on extraction of petroleum ether soluble fats of small birds and mammals. Canadian Journal of Zoology 6, 47 472. Miller, K. & Packard, G.C. (1992) The influence of substrate water potential during incubation on the metabolism of embryonic snapping turtles (Chelydra serpentina). Physiological Zoology 65, 172 187. Morris, K.A., Packard, G.C., Boardman, T.J., Paukstis, G.L. & Packard, M.J. (1983) Effect of the hydric environment on growth of embryonic snapping turtles (Chelydra serpentina). Herpetologica 39, 272 285. Packard, G.C. (1999) Water relations of chelonian eggs and embryos: is wetter better? American Zoologist 39, 289 33. Packard, G.C. & Packard, M.J. (2) Developmental plasticity in Painted Turtles, Chrysemys picta. Functional Ecology 14, 474 483. Packard, G.C., Packard, M.J., Miller, K. & Boardman, T.J. (1987) Influence of moisture, temperature, and substrate on snapping turtle eggs and embryos. Ecology 68, 983 993. Packard, G.C., Packard, M.J., Miller, K. & Boardman, T.J. (1988) Effects of temperature and moisture during incubation on carcass composition of hatchling snapping turtles (Chelydra serpentina). Journal of Comparative Physiology B 158, 117 125. Packard, G.C., Packard, M.J. & Birchard, G.F. (1989) Sexual differentiation and hatching success by painted turtles incubating in different thermal and hydric environments. Herpetologica 45, 385 392. Packard, G.C., Packard, M.J. & Benigan, L. (1991) Sexual differentiation, growth, and hatching success by embryonic painted turtles incubated in wet and dry environments at fluctuating temperatures. Herpetologica 47, 125 132. Packard, G.C., Miller, K. & Packard, M.J. (1993) Environmentally induced variation in body size of turtles hatching in natural nests. Oecologia 93, 445 448. Packard, G.C., Miller, K., Packard, M.J. & Birchard, G.F. (1999) Environmentally induced variation in body size and condition in hatchling snapping turtles (Chelydra serpentina). Canadian Journal of Zoology 77, 278 289. Packard, M.J. & Packard, G.C. (1986) Effect of water balance on growth and calcium mobilization of embryonic painted turtles (Chrysemys picta). Physiological Zoology 59, 398 45. Ratterman, R.J. & Ackerman, R.A. (1989) The water exchange and hydric microclimate of painted turtle (Chrysemys picta) eggs incubating in field nests. Physiological Zoology 62, 159 179. Rowe, J.W., Holy, L., Ballinger, R.E. & Stanley-Samuelson, D. (1995) Lipid provisioning of turtle eggs and hatchlings: total lipid, phospholipid, triacylglycerol and triacylglycerol fatty acids. Comparative Biochemistry and Physiology B 112, 323 33. SAS Institute (199) SAS/STAT User s Guide, Version 6, Vol. 2, 4th edn. SAS Institute Inc., Cary, NC. SAS Institute (1996) SAS/STAT Software: Changes and Enhancements Through Release 6.11. SAS Institute Inc., Cary, NC. Thompson, M.B. (1988) Nest temperatures in the pleurodiran turtle, Emydura macquarii. Copeia 1988, 996 1. Tucker, J.K. (2) Body size and migration of hatchling turtles: inter- and intraspecific comparisons. Journal of Herpetology 34, 541 546. Warren, W.G. (1986) On the presentation of statistical analysis: reason or ritual. Canadian Journal of Forest Research 16, 1185 1191. Wilhoft, D.C. (1986) Eggs and hatchling components of the snapping turtle (Chelydra serpentina). Comparative Biochemistry and Physiology A 84, 483 486. Received 27 November 2; accepted 21 February 21