Using classification trees to study the effects of fisheries management plans on the yield of Merluccius merluccius (Linnaeus, 1758) in the Alboran Sea (Western Mediterranean)

The Common Fisheries Policy (CFP) regulates access to fisheries resources in an attempt to guarantee their conservation, management and exploitation. Direct control of fishing effort is the method that has historically been applied for the management of the Mediterranean fisheries. We used classification trees, a machine learning method, to analyse the effect that different closed seasons have on the yield of European hake (Merluccius merluccius (Linnaeus, 1758)) in “North Alboran Sea”, which corresponds to a geographic subarea of the General Fisheries Commission for the Mediterranean. Various algorithms (CHAID, C4.5, ECHAID, QUEST and J48) were used to generate the classification trees. The study was carried out between 2001 and 2008. During every year fishing was closed for 2 months, which were not necessarily always sequential. Moreover, the months included in the closed season varied between years. When the closed season was in April and May or in April and October, the repercussion on the yield was positive, when compared to other monthly combinations of the closure. The positive effect of the closed season in April and May is more evident than that of April and October. Contrarily, May and June or May and October closed seasons have negative consequences in fishing yields. Finally, a closed season corresponding to March and April showed an intermediate effect between the aforementioned combinations.