Thus, extremely low values of r2 for the calibrations obtained
for C12:0, C17:0, C17:1, C20:0 and C20:1 could be explained by
the similitude of their SD and SEL values reported in Table 1. Higher
coefficients of determination could be achieved by reducing SEL
values, but this approach seems to be unlikely, as cross-validation
errors for the calibration models are quite similar to those obtained
in related studies. However, r2 could be increased by introducing
new samples in the training set in order to enlarge the calibration
range for the fatty acids studied, increasing SD values in this population.
For this purpose, new samples for different campaigns
should be analysed, as changes in animal feeding will vary their
fatty acid profiles.