The strong impact of the type of the statistical analysis chosen on those values was shown in a recent study in gibel carp (Carassius auratus gibeilo) (Wang et al., 2015). Broken-line regression analysis resulted in a clearly lower Met requirement than a polynomial regression analysis. The literature survey in Table 7 shows that those studies using broken-line regression analysis give a lower Met + Cys requirement than those which derived the requirement from polynomial or quadratic regression analyses.