Estimation of regional mortality rates for Lake Erie Walleye Sander vitreus using spatial tag-recovery modeling

Since 1990, Lake Erie Walleyes Sander vitreus have been tagged annually with jaw tags to better understand population dynamics and ecological characteristics of individual spawning populations. Although data collected from this tagging have been used for a variety of purposes, there has been only cursory examination of mortality and movement rates from the recoveries of this long-term tagging program. I used double-tagging and high-reward tagging experiments to estimate tag shedding and reporting rates for jaw-tagged Walleyes. Agency-specific model-averaged estimates of immediate tag retention ranged from 95 to 99%; for chronic shedding, model-averaged instantaneous rates (annual) ranged from 0.07 to 0.28. In general, tag reporting rates were higher for recreational fisheries (range 33--55%) than commercial fisheries (range 10--17%). Reporting rates for both fisheries and tagging basins declined between 1990 and 2000. I then fit a spatial tag-recovery model to jaw-tag recovery data that incorporated tag shedding and reporting rate estimates. As part of this spatial tag-recovery model, I estimated region- and age-specific mortalities and regional movement probabilities. The best performing model based on overdispersion-adjusted Akaike information criteria comparison had age-group-specific movement probabilities, age- and region-specific natural mortalities, and age-group- and region-specific annual fishing mortalities. Commercial fishing mortalities varied considerably during the study, while recreational fishing mortalities were more static. Natural mortalities of age-5 and older Walleyes were lower than those of younger fish in all regions, with natural mortalities ranging from 0.30 to 0.40 for age-4 and younger fish and 0.13 to 0.27 for age-5 and older fish. In Lake Erie's western basin, age-4 natural mortality was lower than that of age-3 fish. To provide guidance with regards to designs of future tagging studies for Lake Erie Walleye, I used stochastic simulations to evaluate bias and precision of fishery and demographic parameter estimates from spatial tag-recovery models under different combinations of high- and low-reward tagging levels, allocations of tags to different age groups, and patterns in age-specific natural mortalities. Parameter precision increased as number of tags released increased and proportion of high-reward tags released increased. Additionally, aside from one exception (i.e., the oldest age group), precision in the estimated mortality rates also increased when I compared a skewed versus balanced tag allocation design. These trends in parameter precision were consistent among different population dynamic scenarios (i.e., assumed natural mortality rates in the data-generating model). The findings presented in this dissertation provide managers with information needed to update the current stock assessment model used to establish safe harvest levels for Lake Erie Walleye as well as insight into how future tagging studies should be conducted to ensure that meaningful fishery and population dynamic parameters can be estimated.