Sensitivity of tag-recovery mortality estimates to inaccuracies in tag shedding, handling mortality, and tag reporting

We used Monte Carlo simulations to evaluate the sensitivity of tag-recovery mortality estimates to inaccuracies in tag shedding, handling mortality, and tag reporting. The data-generating model used in the simulations assumed that tagging was conducted annually for 4 years with tag recoveries occurring over a 4-year period. Several different combinations of instantaneous fishing (F) and natural (M) mortality were evaluated in the simulations. The data-generating model additionally assumed that immediate-shedding and handling-mortality rates equaled 2.5% and 0%, respectively, and that chronic shedding was a sigmoidal function of months since tagging. Two spatial patterns of reporting rates were considered-one where reporting was a function of distance from the tagging site and one where reporting was a random generation across the study area. Maximum likelihood estimates of F and M were calculated from the recovery of tags from the data-generating model under different assumed rates of tag shedding, handling mortality, and tag reporting. We found that assumptions about reporting rates resulted in the most variability in mortality estimates regardless of which combination of F and M was evaluated, with assumptions about chronic shedding also contributing substantially to overall variability in mortality estimates for most mortality combinations. Assumptions about immediate tag shedding and handling mortality had relatively minor effects on mortality estimates compared to reporting rate. When planning a tag-recovery study, care should be taken to ensure that chronic shedding and tag-reporting rates are accurately measured, as inaccurate measurements in these factors can result in significant errors in mortality estimates.