Limited Contribution of Hatchery-Reared Pikeperch to a Wild Population in a Temperate Reservoir: Evidence From Otolith Microchemistry

Abstract

Fish stocking is a common management practice used to mitigate the effects of overfishing and other threats to wild fish populations. It is mainly employed in freshwater ecosystems with complex and variable strategies (different timings, fish lengths and origins), making its effectiveness challenging to assess. To assess and evaluate the prevalence of artificially stocked individuals in a population with a complex management history (pikeperch—Stizostedion lucioperca—from Lipno Reservoir, Czechia), laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) was used to measure the micro-elemental composition at thecore of otoliths to verify the natal origin of fish (wild and stocked individuals) and confirm the origin of fish sources (facilities). Of the ten elements investigated, nine (Ba/Ca, Cu/Ca, K/Ca, Li/Ca, Mg/Ca, Mn/Ca, Na/Ca, Rb/Ca and Sr/Ca) exhibited significant inter-origin variation, whereas Zn/Ca failed to provide a meaningful discriminatory signal. A machine learning classification algorithm, trained with data from 100 fish of known origin, was used to classify 70 fish of unknown origin. The five most relevant variables for correctly classifying fish origins were Rb/Ca, Li/Ca, Sr/Ca, Ba/Ca and Mg/Ca. The model achieved an overall accuracyof 99%, with 97% accuracy for wild fish and 100% for stocked fish. For identifying exact fish sources, overall accuracy for stocking facilities was approximately 89%, with individual facility detection accuracy ranging from 28% to 100%. The model determined that most fish of unknown origin were autochthonous (57 individuals, 81.4%), with a smaller proportion identified as stocked fish (13 individuals, 18.6%). Given the relatively small contribution of stocked fish to the pikeperch population, the efficiency of thestocking programme should be further scrutinised to avoid ineffective management strategies. More broadly, these results highlight that otolith microchemistry combined with machine learning provides a robust framework for evaluating stocking efficiency andsupporting evidence-based fisheries management.