Particle backtracking as a tool to improve stock discrimination capabilities in mixed populations: An example with Lake Erie yellow perch

Background/Question/Methods

Identifying the original source of individuals in mixed populations is a common goal in both aquatic and terrestrial systems. For example, management agencies often seek to quantify the relative contributions of different stocks to their mixed-stock fisheries. Otolith microchemistry and microsatellite data are often employed as tools for this purpose, wherein larvae captured in spawning locations are used to generate natal-site “signatures,” which are used to determine the origins of recruits to the mixed population. This approach assumes that larvae originated at their capture location, which could reduce discrimination capabilities if untrue. Herein, we use geo-referenced individual larval yellow perch (Perca flavescens) hatch date and age data (from otoliths) and water circulation information from western Lake Erie (2006-2007) to demonstrate how a hydrodynamic model can be used in particle “backtracking” mode to improve identification of larval origin and stock discrimination.  

Results/Conclusions

Finding that not all larvae originate in their collection locations, we first show how reassigning larvae to their most probable natal site (based on probabilistic dispersal trajectories) can improve our ability to discriminate among stocks better than the use of otolith microchemistry and genetics alone. For example, larval self-assignment accuracy improved from 60-70% to greater than 90% when using microsatellite data under the most stringent backtracking assignment criterion. When using otolith microchemistry data, backtracking also improved assignment accuracy. Second, we demonstrate that not accounting for dispersal of larvae prior to collection can affect estimates of relative stock contributions to the mixed population. Reassignment after backtracking under the most stringent criterion reduced the rate of failed assignment of juveniles from 38% and 48% to 11% and 4% and substantially altered the estimated relative stock contributions. The results suggest that backtracking may be a useful complement to existing tools for identifying the origin of individuals in mixed populations.