Figure 2 in Growth rates of wild green turtles, Chelonia mydas, at a temperate foraging habitat in the northern Gulf of Mexico: assessing short-term effects of cold-stunning on growth

Figure 2. Graphical summary of generalized additive model fit for somatic growth, in cm straight carapace length (SCL)/year, for St Joseph Bay, Florida conditioned on two growthrate predictors: (A,B) number of previous cold-stunning events and mean carapace length or (C,D) number of previous cold-stunning events and mean condition index. The response variable (growth rate as cm SCL/year) is shown on the y-axis in each panel as a centred scale to ensure valid point-wise 95% credible intervals and comparison between the covariates across the four panels. The width of the mean factor response (number of previous cold-stunning events: A,C) is proportional to sample size with the 95% confidence interval shown by cross bars. Solid curves in B and D are cubic smoothing spline fits for these continuous covariates conditioned on the cofactor (previous cold-stunnings) while the dotted curves in the same panels are point-wise 95% confidence curves around the fits. The data distribution within (B) and (D) is shown by the vertical bars on the topside of the lower x-axis. For instance, (D) shows that most of the data for the mean condition index occur from 1.1 to 1.5 with some extreme outliers. While not statistically significant, it was apparent that expected growth rates were lower for turtles that were exposed to one or two cold-stunning events (A,C). Neither mean size (B) nor mean condition (D) were significant growth-rate predictors for this sample. The sample size (n551) is too small for this study to draw any robust conclusions about the effect of cold-stunning events on juvenile green turtle somatic growth.