Temporal stability of otolith elemental fingerprints discriminates among lagoon nursery habitats

Abstract

The chemical composition of fish otoliths reflects that of the water masses that they inhabit. Otolith elemental compositions can, therefore, be used as natural tags to discriminate among habitats. However, for retrospective habitat identification to be valid and reliable for any adult, irrespective of its age, significant differences in environmental conditions, and therefore otolith signatures, must be temporally stable within each habitat, otherwise connectivity studies have to be carried out by matching year-classes to the corresponding annual fingerprints. This study investigated how various different combinations of chemical elements in otoliths could distinguish, over three separate years, between four coastal lagoon habitats used annually as nurseries by gilthead sea bream (Sparus aurata L.) in the Gulf of Lions (NW Mediterranean). A series of nine elements were measured in otoliths of 301 S. aurata juveniles collected in the four lagoons in 2008, 2010 and 2011. Percentages of correct re-assignment of juveniles to their lagoon of origin were calculated with the Random Forest classification method, considering every possible combination of elements. This revealed both spatial and temporal variations in accuracy of habitat identification, with correct re-assignment to each lagoon ranging from 44 to 99% depending on the year and the lagoon. There were also annual differences in the combination of elements that provided the best discrimination among the lagoons. Despite this, when the data from the three years were pooled, a combination of eight elements (B, Ba, Cu, Li, Mg, Rb, Sr and Y) provided greater than 70% correct re-assignment to each single lagoon, with a multi-annual global accuracy of 79%. When considering the years separately, discrimination accuracy with these elemental fingerprints was above 90% for 2008 and 2010. It decreased to 61% in 2011, when unusually heavy rainfall occurred, which presumably reduced chemical differences among several of the lagoons. This study highlights the need for multi-annual sampling, and multi-elemental analysis, when developing otolith microchemical fingerprints to explore nursery habitat use in coastal fishes.