Selection of the most effective models describing the growth of pike-perch Sander lucioperca larvae in industrial aquaculture

Cultivation of pike-perch during the larval period of development is one of the most complex stages in the technology. Heterogeneity in larval growth and high mortality has hampered the ability to develop or use existing models to describe early pike-perch growth. The solution to the problem of starting feeding of pike perch using artificial feeds made it possible to obtain relatively stable results of cultivation with the possibility of pike-perch growth modeling during the period of early postembryonic development. The aim of the work is to assess the adequacy of existing models capable of describing the growth of pike-perch larvae in industrial aquaculture conditions, as well as to develop and test our own model. Eight nonlinear models used in aquaculture for the mathematical description of growth processes were investigated: three- and four-parameter logistic models, von Bertalanffy, Schnute, Gompertz, Michaelis-Menten and Weibull models, as well as our own experimental linear model. The basis for testing the adequacy of the models was actual data on the growth of pike-perch larvae obtained during their cultivation in tanks entirely on artificial feed. The efficiency and adequacy of the models were determined using the Akaike information criterion AIC, Bayesian information criterion BIC, the mean square error MSE and standard error, as well as the iteration time. The research showed that the growth of pike perch at the larval stage of development is best modeled by four-parameter models: the logistic and experimental linear models, as well as the von Bertalanffy model with three parameters. Moreover, four-parameter models had better AIC, BIC, and R2 coefficient values. While the three-parameter model had the best iteration time, the other models performed significantly worse and, due to unrealistic growth predictions, cannot be used to describe pike-perch larval growth. The results obtained will provide a better understanding of the growth patterns of early pike-perch larvae, help adjust pike-perch rearing technology, and improve farming efficiency.