Analysis of Alburnus tarichi population by machine learning classification methods for sustainable fisheries.

Abstract

An endemic carp species, Alburnus tarichi, inhabits Van Lake Basin in Turkey, where approximately 10,000 tons of this economical and anadromous species are caught each year. Until now, the A. tarichi population has been statistically analyzed based only on caught fish, which provides insufficient information. When these fish reach maturity, do they go to the water sources where their parents spawn, as do salmon? If Alburnus tarichi go to the same locations as their parents to spawn and hatch, then new strains of this species will start forming over time. This study applies two machine learning classification algorithms, k-nearest neighbor (k-NN) and support vector machine (SVM), to an original dataset for A. tarichi population analysis. Fish from nine areas were caught to prepare the original dataset. Five strains were found with machine learning classification algorithms in the Van Lake Basin, and results show that the accuracy levels of the k-NN algorithm were superior to those of the SVM algorithm in the population analysis of Alburnus tarichi.