Application of Machine Learning Methodologies for Unravelling the Philopatry and Dispersal Range of Alosa Species in the Eastern European Atlantic ^†

Abstract

Allis shad, Alosa alosa (Linnaeus, 1758), and twaite shad, Alosa fallax (Lacépède, 1803), are two anadromous fish species belonging to the family Clupeidae. Like all other diadromous species, their populations in European rivers are in decline and their scarcity has become an emerging problem. Anthropogenic activities, mainly large dams and other river obstacles have restricted their range in the eastern Atlantic and western Mediterranean. The studies carried out in Europe in the last 20 years provide a good overview of the conservation status of the shads, however, data on many biological characteristics of their life cycle, especially those corresponding to their development in the marine and coastal environment, are still very scarce. In the Iberian Peninsula, scientific data on their populations are scarcely available. In Galicia, there are studies on the populations of these species in the Ulla and Miño rivers; nevertheless, it is not known whether these species can reside in other Galician rivers suitable for their development. The microchemistry of the otoliths of diadromous fish is strongly correlated with the microchemistry of the water of their river of birth. In the present study, the microchemical composition of the otoliths of 95 adults, distributed between both species, caught at sea and in the Ulla and Miño rivers was analyzed and compared with that of the otoliths of 135 juveniles, distributed between both species, from seven Galician and French rivers. In this way, through this correlation it was possible to quantify the degree of philopatry of the specimens, specify their river of birth and establish their dispersal range during the marine phase. Thus, through the application of Machine Learning methodologies, we provide new information on the dispersal range of these species and the interactions between populations.