Using multinomial models to analyse data from Iberian sardine egg incubation experiments: a comparison with traditional techniques

Multinomial rather than traditional models were applied to improve data analysis of incubation experiments for Iberian sardine (Sardina pilchardus) eggs. Incubation experiments were carried out in 2002 in the Gulf of Cadiz, southwestern Spain, and data were taken from the literature for the Cantabrian Sea area. Rigorous review of the traditional models revealed shortcomings in the statistical framework for modelling the developmental progression of egg stages as well as a lack of transparency of the inherent assumptions. Multinomial model results show that egg stage duration varies for all stages, with stages 3 and 4 having the shortest duration. Comparison of the results between multinomial and traditional models shows that multinornial models provide improved insight into developmental mechanisms than that can be achieved using traditional models.