not-yet-known not-yet-known not-yet-known unknown Results Effect of Multivariate Conditions on Hatchling phenotypes: Body size, fecundity, and egg mass were similar between females from the open versus shaded islands (Supplemental Table 1). Egg survival did not differ between incubation treatments (survival of 87.0% ± 4.56 C.L. in the shaded treatment and 88.8% ± 4.40 C.L. in the open treatment). Incubation period was 7.905 (± 0.402 95% C.I.) days longer for eggs from the shaded treatment than eggs in the open treatment (p < 0.0001) and was negatively correlated with egg mass (β = -0.032 ± 0.009 95% C.I.; p < 0.001). Source island did not affect incubation period (β = 0.23 ± 0.403 95% C.I.; p = 0.269). Incubation treatment influenced hatchling body size. Hatchlings from the shaded treatment were heavier in mass and longer in SVL than hatchlings from the open treatment (Table 1, Figure 4a, b). Hatchling mass and SVL were positively correlated with egg mass; however, an interaction between incubation treatment and egg mass (Table 1) revealed that the positive effect of egg mass (i.e., slope) on SVL was greater for the open vs shaded treatment (Supplemental Figure 3). Neither incubation treatment nor source island substantially influenced hatchling sprint speed (Table 1, Figure 4c). Longer hatchlings ran faster than shorter ones, and the number of stops during a trial had a negative effect on running speed (Table 1). Hatchling water loss was not affected by incubation treatment, source island (Table 1, Figure 4d) or relative humidity during the trial (p = 0.183). Raw means and standard deviations for all phenotypes are in supplemental table 2, and distributions are graphically shown in supplemental figure 4. Family-level variation in reaction norms: Model comparisons revealed that the random intercept model was the best fit for hatchling mass (p < 0.001) and hatchling SVL (p 0.23 and 0.19, respectively (Table 1); however, the null model was best for water loss (p = 1.00) and sprint speed (p = 1.00). No phenotype was best fit with the random slope or random slope and intercept model (Supplemental Table 3). Moreover, interaction terms were not statistically supported in any of the models that directly tested for the genotype by environment interaction, although, like the mixed-model analysis, we observed individual effects of maternal ID and incubation treatment on mass and SVL, but not other traits (Supplemental Table 4). Thus, both the model comparisons and direct tests of G x E interactions indicate family-level variation in body size (but not other traits), and little variation in plasticity (i.e. family-level slopes) for any trait. The analysis of individual reaction norms indicates some variation among family groups in slope and intercept (Figure 5), but variance in slopes did not differ between source islands for any phenotype (Table 2). However, the mean slope for hatchling mass from the open island population was steeper than the mean slope from the shaded island (Table 3, p = 0.0103) but slopes were similar between populations for other traits.