Fig. 1.
Age and population density predict body length in the African cichlid Astatotilapia burtoni. Scatterplots comparing (A) age in days, (B) rearing tank density and (C) body length (estimated by a linear model based on age and rearing tank density; y-axis) with actual measured length (x-axis). Each filled circle represents a fish (n=425; 226 males + 54 females + 145 juveniles). Red dashed lines represent the linear regression of age, density or estimated length on measured length. Pearson correlation (with P-values based on Fisher's z transformation) showed that age alone is a strong positive predictor of body length (A; r=0.8, P=7×10−96), i.e. older fish tend to be larger, and combining age with information on rearing tank density, which correlates negatively with body length (B; r=−0.53, P=3.7×10−32), increases predictive power (C; r=0.86, P=1.2×10−125). A linear model of body length was constructed using the lm() function in R, with age and rearing tank density as independent factors. Coefficients (age=0.1367, density=−0.3736) and intercept (32.1303) values from the model were combined with age and density measurements to estimate body lengths, which are shown in C to correspond with the actual lengths better than age or density alone.