Fig 1.
Phenotype variability in Eurasian coot eggs.
Five example eggs from different clutches. Differences in colour, spottiness, shape and size are very subtle and hardly perceptible by human eye.
Table 1.
List and description of the 27 variables calculated by SpotEgg software to characterise egg phenotype.
Table 2.
Comparison of the accuracy (%) reached by the 12 Support Vector Machines run using a different number of training eggs and sets of explanatory variables.
The percentage of egg classification by random (random accuracy) was 1.1%.
Fig 2.
Variable importance (%) in the best Support Vector Machine (SVM).
Best SVM contained 27 variables and 4 training eggs. Colours group related egg features (ochre: coloration; blue: size and shape; black: spottiness).
Table 3.
Real and simulated Intraclutch Correlation Coefficients (ICC) for egg variables calculated by SpotEgg.