Walleye mortality on Lake of the Woods, Ontario: A joint correction for gear selectivity and spatial sampling bias

Objective Total mortality is a key indicator in fisheries management, and accurate estimation depends on controlling potential sources of sampling bias. Here, we describe an individual-based inverse probability weighting (IPW) approach that corrects for both gear size selectivity and spatially imbalanced sampling effort and apply it to estimate mortality of Walleye Sander vitreus on Lake of the Woods, a culturally and economically important fishery in North America. Methods The IPW method combines size-selectivity coefficients with sampling effort and sector area to weight individual ages according to their relative probability of observation. To properly scale variance estimates, individual weights are normalized so that their sum equals an effective sample size. Total mortality of recruit-sized Walleye (≥350 mm) was estimated from North American standard gill-net survey data collected across six lake sectors using Bayesian methods. Simulations of age segregation and spatial effort imbalance were used to demonstrate how uncorrected spatial imbalance can bias mortality estimates. Results Median total mortality rate was estimated at 0.400/year (95% credible interval = 0.366–0.438) for the combined 2018–2019 catch curve, 0.420/year (0.392–0.449) for the 2021 catch curve and 0.351/year (0.289–0.413) from cohort progression over the study period. Design effects from both area-per-effort weighting and gill-net selectivity were close to 1, indicating that effort per unit area was nearly balanced among sampled sectors and that the recruit-size filter effectively constrained variation in size selectivity in this case study. The spatial-bias simulation showed why correction can still matter: When age segregation and effort imbalance increased across sectors, ignoring area-per-effort weighting produced increasingly biased mortality estimates. Conclusions This study provides corrected mortality estimates for Lake of the Woods Walleye and illustrates how IPW can incorporate known sampling selectivity into mortality analysis. Normalized individual weights allow gear selectivity and spatial sampling imbalance to be combined while keeping uncertainty tied to the effective information content of the sample. Although empirical corrections were small, the simulation showed that ignoring area-per-effort weighting can substantially bias mortality estimates when age structure and sampling effort are both spatially uneven.