Scale-dependent variability in Forsterygion lapillum hatchling otolith chemistry: implications and solutions for studies of population connectivity

Elemental signatures of the otoliths of fishes are increasingly used in connectivity studies to infer the natal origins of settlers or recruits. We evaluated the utility of this approach by assessing variability in trace element signatures within otoliths of hatchlings of the common triplefin Forsterygion lapillum. We sampled hatchling otoliths from eggs collected in a spatially hierarchical design spanning a range of environmental conditions (open coasts, bays and sounds, offshore islands) around Cook Strait, New Zealand. Our results indicate that trace element signatures vary among clutches within sites, among sites within regions and between the North Island and South Island of New Zealand. Because sites within some regions are similar to sites in other regions, we tested a statistical grouping framework based on simulated annealing, which aimed to maximize the power to make robust inferences at a given spatial scale with respect to classification error rate. We further adapted and evaluated a statistical exclusion test framework as an alternative to assignment tests when not all putative source populations could be sampled. For our study system, we found that this exclusion method performed well for some individual sites with sufficiently unique signatures, but did not perform well for groups of sites at larger scales. Overall, our work has highlighted some of the challenges that may limit the utility of hatchling otoliths when used alone for inference of natal origins of fish, and we have presented a set of statistical procedures that may improve the strength of inferences for some ecological questions.